Data processing method and device, electronic equipment and computer readable storage medium

Abstract

The invention provides a data processing method and device, electronic equipment and a computer readable storage medium, and relates to the technical field of data processing, in particular to the technical fields of data comparison, data recall, deep contrast learning, large models and the like. According to the specific implementation scheme, at least one piece of to-be-audited data is acquired from a to-be-audited data set according to at least one preset data processing rule; obtaining at least one piece of standard data from a standard data set according to the at least one data processing rule; and obtaining the similarity between the to-be-audited data and the standard data, and determining an audit result of the to-be-audited data set according to the similarity between the to-be-audited data and the standard data.

Description

Technical Field

[0001] The present disclosure relates to the field of data processing technology, and in particular to the technical fields of data comparison, data recall, deep contrastive learning, large models, etc. Specifically, the present disclosure relates to a data processing method and device, an electronic device, and a computer-readable storage medium. Background Art

[0002] The purpose of an institution’s internal system examination is to verify whether the internal rules, policies and procedures designated by the institution are consistent with applicable laws and regulations and whether they cover all aspects of the laws and regulations. Through internal system examinations, we can protect customer interests, maintain market stability, prevent risks and promote the healthy development of the system.

[0003] The model of relying on manual internal system inspections will inevitably lead to a huge workload and cannot guarantee high coverage. Summary of the invention

[0004] The present disclosure provides a data processing method and device, an electronic device, and a computer-readable storage medium.

[0005] According to a first aspect of the present disclosure, a data processing method is provided, the method comprising:

[0006] Acquire at least one piece of data to be reviewed from the set of data to be reviewed according to at least one preset data processing rule; acquire at least one piece of standard data from the standard set of data according to at least one of the data processing rules;

[0007] The similarity between the data to be reviewed and the standard data is obtained, and the review result of the data set to be reviewed is determined according to the similarity between the data to be reviewed and the standard data.

[0008] According to a second aspect of the present disclosure, there is provided a data processing device, the device comprising:

[0009] A recall module, configured to obtain at least one to-be-audited data from the to-be-audited data set according to at least one preset data processing rule; and to obtain at least one standard data from the standard data set according to at least one of the data processing rules;

[0010] The similarity module is used to obtain the similarity between the data to be reviewed and the standard data, and determine the review result of the data set to be reviewed according to the similarity between the data to be reviewed and the standard data.

[0011] According to a third aspect of the present disclosure, an electronic device is provided, the electronic device comprising:

[0012] at least one processor; and

[0013] A memory in communication with the at least one processor; wherein,

[0014] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the data processing method.

[0015] According to a fourth aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable a computer to execute the above-mentioned data processing method.

[0016] According to a fifth aspect of the present disclosure, a computer program product is provided, comprising a computer program, which implements the above data processing method when executed by a processor.

[0017] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0019] Figure 1 is a flowchart of a data processing method provided by an embodiment of the present disclosure;

[0020] Figure 2 It is a flowchart of some steps of another data processing method provided by an embodiment of the present disclosure;

[0021] Figure 3 It is a flowchart of some steps of another data processing method provided by an embodiment of the present disclosure;

[0022] Figure 4 It is a flowchart of some steps of another data processing method provided by an embodiment of the present disclosure;

[0023] Figure 5 It is a flowchart of some steps of another data processing method provided by an embodiment of the present disclosure;

[0024] Figure 6 is a process diagram of a data processing method for the financial field provided by an embodiment of the present disclosure;

[0025] Figure 7 is a structural schematic diagram of a data processing device provided by an embodiment of the present disclosure;

[0026] Figure 8 It is a block diagram of an electronic device used to implement the data processing method of the embodiment of the present disclosure. DETAILED DESCRIPTION

[0027] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0028] In some related technologies, it is impossible to achieve automated internal system inspections by retaining a professional team of dozens of people and relying on a system management platform to produce content and dynamically maintain internal systems.

[0029] The data processing method and device, electronic device, and computer-readable storage medium provided by the embodiments of the present disclosure are intended to solve at least one of the above technical problems in the prior art.

[0030] The data processing method provided in the embodiments of the present disclosure may be executed by an electronic device such as a terminal device or a server, and the terminal device may be a vehicle-mounted device, a user equipment (UE), a mobile device, a user terminal, a terminal, a cellular phone, a cordless phone, a personal digital assistant (PDA), a handheld device, a computing device, a vehicle-mounted device, a wearable device, etc. The method may be implemented by a processor calling a computer-readable program instruction stored in a memory. Alternatively, the method may be executed by a server.

[0031] Figure 1 FIG. 1 is a flow chart showing a data processing method provided by an embodiment of the present disclosure. Figure 1 As shown in , the method may include step S110 and step S120.

[0032] In step S110, at least one to-be-audited data is obtained from the to-be-audited data set according to at least one preset data processing rule; at least one standard data is obtained from the standard data set according to at least one data processing rule;

[0033] In step S120, the similarity between the data to be reviewed and the standard data is obtained, and the review result of the data set to be reviewed is determined according to the similarity between the data to be reviewed and the standard data.

[0034] For example, in step S110, the preset data processing rule may be one or more.

[0035] In some possible implementations, when there is only one data processing rule, at least one to-be-reviewed data corresponding to the data processing rule is obtained from the to-be-reviewed data set according to the data processing rule; and at least one standard data corresponding to the data processing rule is obtained from the standard data set according to the data processing rule.

[0036] In some possible implementations, when there are multiple data processing rules, for each data processing rule, at least one to-be-reviewed data corresponding to the data processing rule is obtained from the to-be-reviewed data set according to the data processing rule; and at least one standard data corresponding to the data processing rule is obtained from the standard data set according to the data processing rule.

[0037] In some possible implementations, the data set to be reviewed may include multiple data to be reviewed; the standard data set may include multiple standard data; wherein the data to be reviewed may be content formulated with reference to the standard data.

[0038] In some possible implementations, obtaining at least one data to be reviewed corresponding to the data processing rules from the data set to be reviewed according to the data processing rules may be generating a query (query term) according to the data processing rules, and recalling the data to be reviewed that meets the query from multiple data to be reviewed in the data set to be reviewed according to the query.

[0039] Similarly, obtaining at least one standard data corresponding to the data processing rule from the standard data set according to the data processing rule may be generating a query according to the data processing rule, and recalling according to the query to select standard data that meets the query from multiple standard data in the standard data set.

[0040] Among them, the embodiments of the present disclosure do not limit the recall method for recalling, and any method that can achieve recall is within the protection scope of the embodiments of the present disclosure.

[0041] In some possible implementations, the data processing rule may be a recall rule for recalling data, which may be composed of multiple queries.

[0042] In some possible implementations, the to-be-reviewed data set may be a to-be-reviewed document consisting of multiple to-be-reviewed contents; and the standard data set may be a standard document consisting of multiple standard contents. The to-be-reviewed contents may be contents formulated with reference to standard contents.

[0043] In some specific implementations, when there is only one recall rule, at least one to-be-reviewed content corresponding to the recall rule is obtained from the to-be-reviewed document according to the recall rule; and at least one standard content corresponding to the recall rule is obtained from the standard document according to the recall rule.

[0044] In some specific implementations, when there are multiple recall rules, for each recall rule, at least one to-be-reviewed content corresponding to the recall rule is obtained from the to-be-reviewed document according to the recall rule; and at least one standard content corresponding to the recall rule is obtained from the standard document according to the recall rule.

[0045] In some possible implementations, obtaining at least one to-be-reviewed content corresponding to the recall rule from the to-be-reviewed document according to the recall rule may be generating a query (query term) according to the recall rule, and recalling according to the query to select to-be-reviewed content that meets the query from multiple to-be-reviewed contents of the to-be-reviewed document.

[0046] Similarly, obtaining at least one standard content corresponding to the recall rule from the standard document according to the recall rule may be generating a query according to the recall rule, and recalling according to the query to select standard content that meets the query from multiple standard contents of the standard document.

[0047] In some possible implementations, in step S120, obtaining the similarity between the data to be reviewed and the standard data may be a "many-to-many" similarity calculation performed on multiple data to be reviewed and multiple standard data, that is, obtaining the similarity between each data to be reviewed and each standard data, and determining the review result of the data set to be reviewed based on the similarity between each data to be reviewed and each standard data.

[0048] In some possible implementations, the audit result of the to-be-audited data set may include whether the accuracy requirement and the comprehensiveness requirement are met.

[0049] In some specific implementations, when the data set to be reviewed is the document to be reviewed and the standard data set is the standard document, the similarity between the data to be reviewed and the standard data can be obtained by performing a "many-to-many" text content similarity calculation on multiple contents to be reviewed and multiple standard contents, that is, obtaining the similarity between each content to be reviewed and each standard content, and determining the review result of the document to be reviewed based on the similarity between each content to be reviewed and each standard content.

[0050] In some specific implementations, the audit result of the document to be audited may include whether the document to be audited meets the accuracy requirement consistent with the standard document and the comprehensiveness requirement of covering the standard document.

[0051] The embodiments of the present disclosure do not limit the method for calculating the similarity between the data to be reviewed and the standard data. Any method that can achieve similarity calculation is within the protection scope of the embodiments of the present disclosure.

[0052] In the data processing method provided by the embodiment of the present disclosure, the data to be reviewed and the standard data are obtained from the data set to be reviewed and the standard data set respectively according to the preset data processing rules, and the review result of the data to be reviewed is automatically obtained according to the similarity between the data to be reviewed and the standard data set, thereby avoiding the waste of human resources caused by using manpower to review the data set to be reviewed according to the standard data set. At the same time, since the data to be reviewed and the standard data are obtained according to the data processing rules, it is possible to obtain part of the content of the data set to be reviewed and review this part of the content by setting the data processing rules.

[0053] The data processing method of the embodiment of the present disclosure is described in detail below.

[0054] In some possible implementations, the data set to be reviewed in the method provided by the embodiments of the present disclosure may be a collection of internal systems formulated by an organization, specifically, the documents that constitute the internal systems; the standard data set may be a collection of laws and regulations that the organization uses as reference when formulating its internal systems, specifically, the legal and regulatory documents that the internal systems use as reference.

[0055] Therefore, the method provided by the embodiment of the present disclosure can be used to review internal systems to review whether the internal systems conflict with laws and regulations, or whether the internal systems cover all aspects of laws and regulations.

[0056] In some possible implementations, the data set to be reviewed may include multiple clauses to be reviewed (specifically, they may be internal system clauses formulated by the organization); the standard data set may include multiple standard clauses (specifically, they may be legal and regulatory clauses that the organization uses as reference when formulating its internal system).

[0057] In some possible implementations, the data to be reviewed may include at least one clause to be reviewed; and the standard data may include at least one standard clause.

[0058] In some possible implementations, obtaining at least one to-be-reviewed data from the to-be-reviewed data set according to at least one data processing rule may be obtaining, according to the data processing rule, a to-be-reviewed clause that complies with the data processing rule from multiple to-be-reviewed clauses included in the to-be-reviewed data set.

[0059] In some possible implementations, obtaining at least one standard data from a standard data set according to at least one data processing rule may be obtaining a standard clause that complies with the data processing rule from a plurality of standard clauses included in the standard data set according to the data processing rule.

[0060] In some possible implementations, the BM25 (Best Match 25, best match of the 25th algorithm iteration) algorithm can be used to obtain at least one to-be-reviewed data corresponding to the data processing rule from the to-be-reviewed data set according to the data processing rule, and to obtain at least one standard data corresponding to the data processing rule from the standard data set according to the data processing rule.

[0061] BM25 is an information retrieval algorithm used to calculate the match between the target and the query. It evaluates the match between the target content and the query based on the term frequency (TF) of the text in the target and the term frequency of the query, as well as the inverse document frequency (IDF) of the terms in the target.

[0062] The core idea of ​​the BM25 algorithm is to determine the relevance of the target content to the query by calculating the score of the query terms.

[0063] In some possible implementations, the BM25 algorithm can be used to obtain the pending clauses that comply with the data processing rules from the multiple pending clauses included in the pending dataset according to the data processing rules. Specifically, calculating whether the pending clauses of the pending dataset comply with the data processing rules may include the following steps:

[0064] Preprocess documents and queries: Segment the terms to be reviewed in the review data set, remove insignificant terms such as stop words, generate query terms according to data processing rules, and obtain the terms corresponding to the query terms.

[0065] Calculate the inverse document frequency (IDF): For each term, calculate its inverse document frequency in the terms to be reviewed. IDF is usually calculated using the following formula:

[0066]

[0067] Where W represents a term, N represents the number of pending clauses in the pending-review data set, and n represents the number of pending clauses containing the term W.

[0068] Calculate term frequency (TF): For each pending term and query term, calculate the term frequency of each term in the pending term or query term.

[0069] Calculate the matching score: For each item to be reviewed, calculate its matching score with the query. The matching score is calculated as follows:

[0070]

[0071] Where D represents the pending terms, Q represents the query term, q represents the term in the query term, TF(q,D) represents the term frequency of the term in the pending terms, IDF(q) represents the inverse document frequency of the term, k1 and K are adjustable parameters used to control the normalization of TF.

[0072] BM25 introduces a parameter k1, which can be used to control the saturation of the frequency of terms in the document. This means that BM25 can adjust parameters according to different data sets and requirements to adapt to different text information retrieval tasks. BM25 uses dynamic vocabulary weights when calculating the relevance between the terms to be reviewed and the query, which means that BM25 better considers the relationship between the query terms and other terms in the terms to be reviewed, improving the accuracy of retrieval.

[0073] At the same time, in the application scenario corresponding to the method provided in the embodiment of the present disclosure, the query term may not appear frequently in the terms to be reviewed, and BM25 improves the accuracy of the retrieval of the terms to be reviewed by using IDF (inverse document frequency) to assign lower weights to rare words.

[0074] In some possible implementations, obtaining at least one standard data from a standard data set according to at least one data processing rule may be obtaining a standard clause that complies with the data processing rule from a plurality of standard clauses included in the standard data set according to the data processing rule.

[0075] In some possible implementations, the BM25 algorithm can be used to obtain standard clauses that comply with the data processing rules from multiple standard clauses included in the standard data set according to the data processing rules. Specifically, the specific process of calculating whether the standard clauses of the standard data set comply with the data processing rules is the same as the specific process of calculating whether the to-be-reviewed clauses of the to-be-reviewed data set comply with the data processing rules, which will not be repeated here.

[0076] In some possible implementations, when the data to be reviewed includes at least one clause to be reviewed and the standard data includes at least one standard clause, the review result of the data set to be reviewed includes an accuracy review result and a comprehensiveness review result.

[0077] Figure 2 The flowchart shows a method for determining the audit result of the data set to be audited based on the similarity between the data to be audited and the standard data when the audit result of the data set to be audited includes the accuracy audit result and the comprehensive audit result. Figure 2 As shown in , determining the audit result of the data set to be audited according to the similarity between the data to be audited and the standard data may include step S210, step S220, and step S230.

[0078] In step S210, the similarity between the clause to be reviewed and the standard clause is obtained;

[0079] In step S220, if there is at least one standard clause in the at least one standard clause whose similarity with the clause to be reviewed meets the preset threshold requirement, it is determined that the clause to be reviewed meets the accuracy requirement;

[0080] In step S230, when at least one of the at least one clause to be reviewed has a similarity with the standard clause that meets a preset threshold requirement, it is determined that the clause to be reviewed meets the comprehensiveness requirement.

[0081] In some possible implementations, in step S210, obtaining the similarity between the clause to be reviewed and the standard clause may be obtaining the similarity between each clause to be reviewed and each standard clause.

[0082] In some possible implementations, a large model may be used to obtain the similarity between the data to be reviewed and the standard data.

[0083] Among them, the big model refers to a machine learning model with a large number of parameters and complex structure. It is a model based on machine learning and natural language processing technology. Specifically, it learns the ability to serve human language understanding and generation by training on large amounts of text data.

[0084] In the data processing method provided in the embodiment of the present disclosure, a large model that can obtain text features of the input text can be obtained through training with a large amount of text data, and the similarity between the data to be reviewed and the standard data input into the large model can be determined based on the similarity of the text features output by the large model.

[0085] Large models have strong capabilities for processing large amounts of data, handling complex problems, and have higher accuracy and performance. Large models have more parameters and more complex structures, and can more accurately express data distribution and learn more complex features, thereby improving the accuracy and performance of the model. Therefore, using large models to obtain the similarity between the data to be reviewed and the standard data can improve the accuracy of the obtained similarity.

[0086] In some specific implementations, deep comparative learning based on SimCSE (unsupervised comparative sentence vector representation) can be used to obtain the similarity between the data to be reviewed and the standard data.

[0087] In some specific implementations, the terms to be reviewed and the standard terms can be input into the pre-trained SimCSE to obtain their corresponding text vectors, and then the similarity of the two text vectors can be calculated through the Cosine-Sim function to obtain the similarity between the terms to be reviewed and the standard terms.

[0088] Specifically, the clauses to be reviewed and the standard clauses can be input into the trained SimCSE respectively to obtain the features corresponding to the clauses to be reviewed and the features corresponding to the standard clauses output by SimCSE, and the features corresponding to the clauses to be reviewed and the features corresponding to the standard clauses can be input into the Pooling layer respectively to obtain the text vectors corresponding to the clauses to be reviewed and the text vectors corresponding to the standard clauses, and the Cosine-Sim function can be used to calculate the similarity between the text vectors corresponding to the clauses to be reviewed and the text vectors corresponding to the standard clauses, and the calculated similarity can be used as the similarity between the clauses to be reviewed and the standard clauses.

[0089] Among them, the training process of SimCSE may include: inputting the same training data (specifically, legal and regulatory sentences) into the model including SimCSE twice to obtain feature vectors corresponding to two different training data output by the model; taking the two feature vectors of the same training data in the model as positive examples, and taking the output of the model corresponding to different training data as negative examples, changing the model parameters of SimCSE to optimize the contrast loss, increasing the similarity between positive examples, and reducing the similarity between negative examples to train SimCSE.

[0090] Since the terms to be reviewed and the standard terms are essentially sentences composed of text, the accuracy of the similarity results obtained by obtaining the similarity between the data to be reviewed and the standard data by using deep comparative learning based on SimCSE can be increased.

[0091] In some possible implementations, in step S220, for each clause to be reviewed, only if there is a standard clause in the standard clauses whose similarity with the clause to be reviewed meets a preset threshold, is it determined that the clause to be reviewed meets the accuracy requirement.

[0092] If there is at least one standard clause in the standard clauses whose similarity with the clause to be reviewed meets the preset threshold, it means that the clause to be reviewed meets the requirements of these standard clauses, that is, the clause to be reviewed may be formulated based on these standard clauses and meets the requirements of these standard clauses.

[0093] In some specific implementations, when the data set to be reviewed is a collection of internal systems formulated by an organization; and the standard data set is a collection of laws and regulations that the organization uses as reference when formulating its internal systems, there is at least one standard clause in the standard clauses whose similarity with the clause to be reviewed meets a preset threshold, indicating that the internal system clause meets the provisions of the referenced legal and regulatory clauses, and thus meets the accuracy requirements.

[0094] In some possible implementations, in step S230, for each standard clause, only if there is a clause to be reviewed among the clauses to be reviewed whose similarity with the standard clause meets a preset threshold, is it determined that the standard clause meets the accuracy requirement.

[0095] If there is at least one clause in the clauses to be reviewed whose similarity with the standard clause meets the preset threshold, it means that the standard clause has a corresponding clause to be reviewed. If all standard clauses have corresponding clauses to be reviewed, it means that the clauses to be reviewed cover all standard clauses and meet the comprehensiveness requirement.

[0096] In some specific implementations, when the data set to be reviewed is a collection of internal systems formulated by an organization; and the standard data set is a collection of laws and regulations for reference in formulating internal systems by an organization, there is at least one clause to be reviewed in the clauses to be reviewed whose similarity with the standard clause meets a preset threshold, indicating that the legal and regulatory clause has a corresponding internal system clause corresponding to it, that is, the laws and regulations have been formed into internal system clauses. When all laws and regulations have been formed into internal system clauses, the internal system clauses meet the comprehensiveness requirements.

[0097] Accuracy and comprehensiveness are important indicators for measuring the formulation of laws and regulations. The above method can be used to simply calculate the accuracy and comprehensiveness of systems or legal and regulatory documents.

[0098] In some possible implementations, when there are multiple preset data processing rules, for each preset data processing rule, the data to be reviewed and the standard data are obtained according to the data processing rule, and the review result of the data set to be reviewed is obtained according to the similarity between the data to be reviewed and the standard data.

[0099] Figure 3 A flowchart of an implementation method of obtaining the data to be reviewed and the standard data according to each preset data processing rule, and obtaining the review result of the data set to be reviewed according to the similarity between the data to be reviewed and the standard data is shown, as shown in FIG. Figure 3 As shown, it may include step S310 and step S320.

[0100] In step S310, at least one to-be-audited data corresponding to the data processing rule is obtained from the to-be-audited data set according to the data processing rule; at least one standard data corresponding to the data processing rule is obtained from the standard data set according to the data processing rule;

[0101] In step S320, the similarity between the data to be reviewed corresponding to the data processing rules and the standard data corresponding to the data processing rules is obtained, and the review result of the data set to be reviewed corresponding to the data processing rules is determined according to the similarity between the data to be reviewed corresponding to the data processing rules and the standard data corresponding to the data processing rules.

[0102] In some possible implementations, in step S310, the data to be reviewed and the standard data obtained according to a data processing rule are the standard data and the data to be reviewed corresponding to the data processing rule.

[0103] The specific method for obtaining the data to be reviewed and the standard data according to the data processing rules has been described above and will not be repeated here.

[0104] In some possible implementations, in step S320, the similarity between each data to be reviewed and each standard data corresponding to the data processing rule is calculated, and the accuracy of the data set to be reviewed under the data processing rule is determined based on whether the similarity between the data to be reviewed corresponding to each data processing rule and the standard data corresponding to the data processing rule meets a preset threshold; the comprehensiveness of the data set to be reviewed under the data processing rule is determined based on whether the similarity between the standard data corresponding to each data processing rule and the data to be reviewed corresponding to the data processing rule meets a preset threshold.

[0105] The specific method for calculating the similarity between each to-be-audited data and each standard data corresponding to the data processing rule is as described above and will not be repeated here.

[0106] In some specific implementations, when the data processing rule is a recall rule, the data set to be reviewed is a document to be reviewed, and the standard data set is a standard document, the content to be reviewed and the standard content obtained according to a recall rule are the standard content and content to be reviewed corresponding to the recall rule.

[0107] In some specific implementations, the similarity between each content to be reviewed and each standard content corresponding to the recall rule is calculated, and the accuracy of the document to be reviewed under the recall rule is determined based on whether the similarity between the content to be reviewed corresponding to each recall rule and the standard content corresponding to the recall rule meets the preset threshold; the comprehensiveness of the document to be reviewed under the recall rule is determined based on whether the standard content corresponding to each recall rule and the content to be reviewed corresponding to the recall rule meet the preset threshold.

[0108] The data to be reviewed and the standard data corresponding to the data processing rules obtained according to the data processing rules are contents related to the data processing rules. The accuracy and comprehensiveness of the contents related to the data processing rules in the data set to be reviewed are determined by calculating the similarity between the data to be reviewed and the standard data corresponding to the data processing rules.

[0109] That is to say, the method provided by the embodiment of the present disclosure can calculate the accuracy and comprehensiveness of part of the content of a data set. Compared with calculating the accuracy and comprehensiveness of the entire data set, it requires less calculation and can eliminate unimportant data in the data set, reducing unnecessary calculations.

[0110] In some possible implementations, the to-be-audited data set and the standard data set may be obtained by preprocessing documents.

[0111] Figure 4 A schematic diagram of a process for obtaining a dataset to be reviewed and a standard dataset is shown. Figure 4 As shown, it may include step S410 and step S420.

[0112] In step S410, the format and layout of the document to be parsed and reviewed are parsed to obtain a data set to be reviewed;

[0113] In step S420, format analysis and layout analysis are performed on the standard document to be analyzed to obtain a standard data set.

[0114] In some possible implementations, in step S410 and step S420, the audit document to be parsed and the standard document to be parsed may be the original internal system document and the original legal and regulatory document, i.e., the internal system document for personnel to view, which also has a format, layout, etc. to facilitate personnel to view. Specifically, it may be in the form of a computer-readable document or in the form of a picture.

[0115] In some possible implementations, the doc-parser service can be used to perform format and layout analysis on the document, achieve element restoration, position restoration, content restoration, and sequence restoration, and obtain a document with no format and no layout, but only content. For example, a data set containing only the clauses of internal systems and laws and regulations can be obtained.

[0116] Through format parsing and layout parsing, unformatted documents can be obtained, avoiding the impact of document format on subsequent data processing and improving the accuracy of the review.

[0117] In some possible implementations, after obtaining the dataset to be reviewed and the standard dataset, the contents of the dataset to be reviewed and the standard dataset may be inserted into corresponding indexes to facilitate the subsequent recall process according to data processing rules.

[0118] Figure 5 The figure shows a flow chart of an implementation of inserting the contents of the to-be-reviewed dataset and the standard dataset into the corresponding index after obtaining the to-be-reviewed dataset and the standard dataset. Figure 4 As shown in , step S510 may be included.

[0119] In step S510, the dataset to be reviewed and the standard dataset are inserted into corresponding indexes, and the document attributes of the documents to which the dataset to be reviewed and the standard dataset belong are saved.

[0120] In some possible implementations, pending review and standard indexes are established in ES (Elasticsearch, a full-text search engine), and the contents of the pending review data set and the standard data set obtained after parsing are inserted into the corresponding pending review and standard indexes. At the same time, the document attributes can be stored together with the data set and inserted into the corresponding index.

[0121] In some specific implementations, when the data set to be reviewed includes internal system documents formulated by an organization, and the standard data set includes legal and regulatory documents for reference when the organization formulates its internal systems, the data set to be reviewed obtained is a data set including multiple internal system clauses, and the standard data set is a data set including multiple legal and regulatory clauses.

[0122] Establish internal and external regulation indexes in ES, and insert the unreviewed dataset and standard dataset obtained after parsing into the index at the clause granularity. That is, insert each internal system clause obtained after parsing into the internal regulation index, and write the internal system to which the internal clause belongs together; insert each legal and regulatory clause obtained after parsing into the external regulation index, and write the laws and regulations to which the legal and regulatory clause belongs together.

[0123] In some possible implementations, the method provided by the embodiments of the present disclosure may be applied to the financial field. Specifically, the method provided by the embodiments of the present disclosure may be applied to the internal system inspection of financial institutions.

[0124] Financial institutions need to meet the following two requirements when formulating internal systems:

[0125] 1. The internal systems of financial institutions must be consistent with applicable financial regulations.

[0126] This means that the internal system should be consistent with the requirements set out in the regulations, including the purpose, principles and specific provisions of the regulations. Institutions need to study the regulations carefully and ensure that their internal systems cover what is required in the regulations. In addition, institutions should promptly adjust their internal systems to adapt to changes in regulations to ensure continued consistency with the regulations.

[0127] 2. The internal systems of financial institutions should be comprehensive and cover all aspects of laws and regulations.

[0128] Financial regulations are usually multi-dimensional and multi-layered, and include many provisions and requirements. Therefore, the internal systems of financial institutions should comprehensively cover these regulations and requirements to ensure that no important content is omitted. Institutions need to fully interpret and analyze the regulations and formulate corresponding internal systems, covering all aspects from business processes to internal controls, risk management, reporting and disclosure, to ensure that the institution complies with regulatory requirements in all key areas.

[0129] Therefore, when the data set to be audited is the internal system formulated by the financial institution and the standard data set is the laws and regulations used as reference by the financial institution in formulating its internal system, the audit results of the data set to be audited include accuracy audit results and comprehensiveness audit results.

[0130] In some possible implementations, at least one data processing rule includes at least one of compliance requirement provisions, prohibition provisions, frequency provisions, deadline provisions, and amount provisions.

[0131] In some possible implementations, compliance requirements provisions (the following conditions 'or' are met) may include: all "should" screening; compliance requirements of "should + principle" removed from complete sentences in laws and regulations; all "should + establish" screening; all "should + possess" screening; all "should + review" screening; all "should + filing" screening; all "should + report" screening; all "should + mechanism" screening; all "should + fund custody business" compliance requirements.

[0132] In some possible implementations, prohibitive provisions (the following conditions 'or' are met) may include: all "shall not, prohibited, not compliant, not in accordance with, violate, violate + fund custody business" screening.

[0133] In some possible implementations, the frequency requirements (the following conditions are met "or") include: all "annual + timely submission + fund custody business" screening; all "annual + reporting + fund custody business" screening.

[0134] In some possible implementations, the time limit regulations may include: all “intra-day, intra-month, intra-year + fund custody business” screening.

[0135] In some possible implementations, the amount regulations may include: all "10,000 yuan, 100 million yuan, minimum limit + fund custody business" screening.

[0136] By setting up data processing rules, it is possible to automatically check the compliance requirements, prohibitions, frequency requirements, deadlines and amounts of the internal systems formulated by financial institutions, thereby improving the efficiency of the internal control part of financial institutions and effectively reducing labor costs.

[0137] Figure 6 A schematic diagram showing a specific process of applying the data processing method provided by an embodiment of the present disclosure to the internal system physical examination of a financial institution.

[0138] like Figure 6As shown, the internal regulation documents belonging to the internal system of the financial institution are taken as the review documents to be parsed, and the external regulation documents belonging to the legal and regulatory documents used as references by the financial institution in formulating the internal system are taken as the standard documents to be parsed. They are input into the doc-parser service (this service provides standardized "format parsing" and "layout parsing" capabilities, which can achieve element restoration, position restoration, content restoration, and sequence restoration, and supports complex layouts such as multiple columns and mixed types), and obtain the list of clauses of the internal system and regulations and the list of clauses of the laws and regulations.

[0139] Establish internal and external regulation indexes in ES, and use the parsed clause list as the content of the data to be reviewed and the standard data set at the clause granularity to form the data set to be reviewed and the standard data set, and insert them into the corresponding internal or external regulation index, and save the name of the regulation to which each clause belongs.

[0140] Then, according to the above data processing rules, use the BM25 algorithm to retrieve qualified internal and external regulation clauses (i.e. data to be reviewed and standard data) from the internal and external regulation indexes of ES respectively.

[0141] A "many-to-many" similarity calculation is performed on the recalled internal and external regulation clauses of each data processing rule, and a similarity threshold is set to determine whether a recalled internal regulation clause has a corresponding external regulation clause to verify the consistency of the internal and external regulations under the data processing rule. Conversely, the comprehensiveness of the internal regulation under the data processing rule can be verified.

[0142] Based on Figure 1 The same principle as shown in the method, Figure 7 A schematic diagram of the structure of a data processing device provided by an embodiment of the present disclosure is shown. Figure 7 As shown, the data processing device 70 may include:

[0143] The recall module 710 is used to obtain at least one to-be-reviewed data from the to-be-reviewed data set according to at least one preset data processing rule; and obtain at least one standard data from the standard data set according to at least one data processing rule;

[0144] The similarity module 720 is used to obtain the similarity between the data to be reviewed and the standard data, and determine the review result of the data set to be reviewed according to the similarity between the data to be reviewed and the standard data.

[0145] In the data processing device provided by the embodiment of the present disclosure, the data to be reviewed and the standard data are obtained from the data set to be reviewed and the standard data set respectively according to the preset data processing rules, and the review result of the data to be reviewed is automatically obtained according to the similarity between the data to be reviewed and the standard data set, thereby avoiding the waste of human resources caused by using manpower to review the data set to be reviewed according to the standard data set. At the same time, since the data to be reviewed and the standard data are obtained according to the data processing rules, part of the content of the data set to be reviewed can be obtained by setting the data processing rules, and this part of the content can be reviewed.

[0146] In some possible implementations, the data to be reviewed includes at least one clause to be reviewed; the standard data includes at least one standard clause; the similarity module 720 includes: a similarity unit, used to obtain the similarity between the clause to be reviewed and the standard clause; an accuracy unit, used to determine that the clause to be reviewed meets the accuracy requirement when at least one standard clause in at least one standard clause has a similarity with the clause to be reviewed that meets the preset threshold requirement; a comprehensiveness unit, used to determine that the clause to be reviewed meets the comprehensiveness requirement when at least one standard clause in at least one standard clause has a similarity with the clause to be reviewed that meets the preset threshold requirement.

[0147] In some possible implementations, when there are multiple preset data processing rules, for each preset data processing rule, the recall module 710 is used to obtain at least one pending data corresponding to the data processing rule from the pending data set according to the data processing rule; obtain at least one standard data corresponding to the data processing rule from the standard data set according to the data processing rule; the similarity module 720 is used to obtain the similarity between the pending data corresponding to the data processing rule and the standard data corresponding to the data processing rule, and determine the audit result of the pending data set corresponding to the data processing rule according to the similarity between the pending data corresponding to the data processing rule and the standard data corresponding to the data processing rule.

[0148] In some possible implementations, the recall module 710 is used to: use the 25th algorithm iteration of the best matching algorithm to obtain at least one data to be reviewed from the data set to be reviewed according to at least one preset data processing rule; use the 25th algorithm iteration of the best matching algorithm to obtain at least one standard data from the standard data set according to at least one data processing rule.

[0149] In some possible implementations, the similarity module 720 is used to obtain the similarity between the to-be-audited data and the standard data using deep contrastive learning based on unsupervised contrast sentence vector representation.

[0150] In some possible implementations, the data processing device 70 also includes: a first parsing module, used to perform format parsing and layout parsing on the audit document to be parsed, and obtain the data set to be audited; a second parsing module, used to perform format parsing and layout parsing on the standard document to be parsed, and obtain the standard data set.

[0151] In some possible implementations, the data processing device 70 further includes: an indexing module, configured to insert the to-be-audited data set and the standard data set into corresponding indexes, and save document properties of documents to which the to-be-audited data set and the standard data set belong.

[0152] In some possible implementations, the dataset to be reviewed includes internal systems formulated by the organization; the standard dataset includes laws and regulations that the organization uses as reference when formulating its internal systems.

[0153] In some possible implementations, the institution is a financial institution; the data processing rules include at least one of compliance requirement regulations, prohibition regulations, frequency regulations, time limit regulations, and amount regulations.

[0154] It can be understood that the above modules of the data processing device in the embodiment of the present disclosure have the function of implementing Figure 1 The functions of the corresponding steps of the data processing method in the embodiment shown in . The function can be implemented by hardware, or by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the above functions. The above modules can be software and / or hardware, and the above modules can be implemented separately or integrated with multiple modules. For the functional description of each module of the above data processing device, please refer to Figure 1 The corresponding description of the data processing method in the embodiment shown in will not be repeated here.

[0155] In the technical solution disclosed herein, the acquisition, storage and application of user personal information involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0156] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0157] The electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the data processing method provided in the embodiment of the present disclosure.

[0158] Compared with the prior art, the electronic device obtains the data to be reviewed and the standard data from the data set to be reviewed and the standard data set respectively according to the preset data processing rules, and automatically obtains the review result of the data to be reviewed according to the similarity between the data to be reviewed and the standard data set, thereby avoiding the waste of human resources caused by using manpower to review the data set to be reviewed according to the standard data set. At the same time, since the data to be reviewed and the standard data are obtained according to the data processing rules, part of the content of the data set to be reviewed can be obtained by setting the data processing rules, and this part of the content can be reviewed.

[0159] The readable storage medium is a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable a computer to execute a data processing method as provided in an embodiment of the present disclosure.

[0160] Compared with the prior art, the readable storage medium obtains the data to be reviewed and the standard data from the data set to be reviewed and the standard data set respectively according to the preset data processing rules, and automatically obtains the review result of the data to be reviewed according to the similarity between the data to be reviewed and the standard data set, thereby avoiding the waste of human resources caused by using manpower to review the data set to be reviewed according to the standard data set. At the same time, since the data to be reviewed and the standard data are obtained according to the data processing rules, part of the content of the data set to be reviewed can be obtained by setting the data processing rules, and this part of the content can be reviewed.

[0161] The computer program product includes a computer program, and when the computer program is executed by a processor, the computer program implements the data processing method provided in the embodiment of the present disclosure.

[0162] Compared with the prior art, the computer program product obtains the data to be reviewed and the standard data from the data set to be reviewed and the standard data set respectively according to the preset data processing rules, and automatically obtains the review result of the data to be reviewed according to the similarity between the data to be reviewed and the standard data set, thereby avoiding the waste of human resources caused by using manpower to review the data set to be reviewed according to the standard data set. At the same time, since the data to be reviewed and the standard data are obtained according to the data processing rules, part of the content of the data set to be reviewed can be obtained by setting the data processing rules, and this part of the content can be reviewed.

[0163] Figure 8A schematic block diagram of an example electronic device 600 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0164] like Figure 8 As shown, the device 800 includes a computing unit 801, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 into a random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the device 800 can also be stored. The computing unit 801, the ROM 802, and the RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0165] A number of components in the device 800 are connected to the I / O interface 805, including: an input unit 806, such as a keyboard, a mouse, etc.; an output unit 807, such as various types of displays, speakers, etc.; a storage unit 808, such as a disk, an optical disk, etc.; and a communication unit 809, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 809 allows the device 800 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0166] The computing unit 801 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 801 performs the various methods and processes described above, such as data processing methods. For example, in some embodiments, the data processing method may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 808. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 800 via ROM 802 and / or communication unit 809. When the computer program is loaded into RAM 803 and executed by the computing unit 801, one or more steps of the data processing method described above may be performed. Alternatively, in other embodiments, the computing unit 801 may be configured to perform the data processing method in any other appropriate manner (e.g., by means of firmware).

[0167] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0168] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0169] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0170] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0171] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0172] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0173] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0174] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A data processing method, comprising: Acquire at least one piece of data to be reviewed from the set of data to be reviewed according to at least one preset data processing rule; acquiring at least one standard data from the standard data set according to at least one of the data processing rules; The similarity between the data to be reviewed and the standard data is obtained, and the review result of the data set to be reviewed is determined according to the similarity between the data to be reviewed and the standard data.

2. The method according to claim 1, wherein: The data to be reviewed includes at least one clause to be reviewed; the standard data includes at least one standard clause; The obtaining of the similarity between the data to be reviewed and the standard data, and determining the review result of the data set to be reviewed according to the similarity between the data to be reviewed and the standard data, includes: Obtaining the similarity between the clause to be reviewed and the standard clause; In the case where at least one of the standard clauses has a similarity with the clause to be reviewed that meets a preset threshold requirement, determining that the clause to be reviewed meets the accuracy requirement; In the case where at least one of the clauses to be reviewed has a degree of similarity between the clause to be reviewed and the standard clause that meets a preset threshold requirement, it is determined that the clause to be reviewed meets the comprehensiveness requirement.

3. The method according to claim 1, wherein: In the case where there are multiple preset data processing rules, for each preset data processing rule, The step of obtaining at least one piece of data to be reviewed from the set of data to be reviewed according to at least one preset data processing rule; Acquiring at least one standard data from the standard data set according to at least one of the data processing rules comprises: According to the data processing rule, at least one to-be-audited data corresponding to the data processing rule is obtained from the to-be-audited data set; according to the data processing rule, at least one standard data corresponding to the data processing rule is obtained from the standard data set; The obtaining of the similarity between the data to be reviewed and the standard data, and determining the review result of the data set to be reviewed according to the similarity between the data to be reviewed and the standard data, includes: Obtain the similarity between the data to be reviewed corresponding to the data processing rule and the standard data corresponding to the data processing rule, and determine the review result of the data set to be reviewed corresponding to the data processing rule according to the similarity between the data to be reviewed corresponding to the data processing rule and the standard data corresponding to the data processing rule.

4. The method according to claim 1, wherein: The step of obtaining at least one piece of data to be reviewed from the set of data to be reviewed according to at least one preset data processing rule; Acquiring at least one standard data from the standard data set according to at least one of the data processing rules comprises: Using the 25th algorithm iteration best matching algorithm to obtain at least one data to be reviewed from the data set to be reviewed according to at least one preset data processing rule; The best matching algorithm is used for the 25th algorithm iteration to obtain at least one standard data from the standard data set according to at least one of the data processing rules.

5. The method according to claim 1, wherein: The obtaining of the similarity between the data to be reviewed and the standard data includes: The similarity between the data to be reviewed and the standard data is obtained by using deep contrastive learning based on unsupervised contrast sentence vector representation.

6. The method according to claim 1, wherein: The step of obtaining at least one piece of data to be reviewed from the set of data to be reviewed according to at least one preset data processing rule; Before acquiring at least one standard data from the standard data set according to at least one of the data processing rules, the method further includes: Perform format and layout analysis on the documents to be reviewed and obtain the data set to be reviewed; Perform format and layout analysis on the standard document to be analyzed to obtain a standard data set.

7. The method according to claim 6, wherein: The format parsing and layout parsing of the audit document to be parsed are performed to obtain the audit data set; the format parsing and layout parsing of the standard document to be parsed are performed to obtain the standard data set, and the following is further included: The to-be-reviewed data set and the standard data set are inserted into corresponding indexes, and document properties of the documents to which the to-be-reviewed data set and the standard data set belong are saved.

8. The method according to claim 1, wherein: The data set to be reviewed includes the internal systems formulated by the institution; the standard data set includes the laws and regulations that the institution uses as reference in formulating its internal systems.

9. The method according to claim 8, wherein: The institution is a financial institution; the data processing rules include at least one of compliance requirements, prohibition regulations, frequency regulations, time limit regulations, and amount regulations.

10. A data processing device, comprising: A recall module, used for obtaining at least one to-be-reviewed data from the to-be-reviewed data set according to at least one preset data processing rule; acquiring at least one standard data from the standard data set according to at least one of the data processing rules; The similarity module is used to obtain the similarity between the data to be reviewed and the standard data, and determine the review result of the data set to be reviewed according to the similarity between the data to be reviewed and the standard data.

11. The device according to claim 10, wherein: The data to be reviewed includes at least one clause to be reviewed; the standard data includes at least one standard clause; The similarity module comprises: A similarity unit, used to obtain the similarity between the clause to be reviewed and the standard clause; an accuracy unit, configured to determine that the clause to be reviewed meets the accuracy requirement when at least one of the standard clauses has a similarity with the clause to be reviewed that meets a preset threshold requirement; The comprehensiveness unit is used to determine that the clause to be reviewed meets the comprehensiveness requirement when at least one of the clauses to be reviewed has a similarity with the standard clause that meets a preset threshold requirement.

12. The device according to claim 10, wherein: In the case where there are multiple preset data processing rules, for each preset data processing rule, The recall module is used to obtain at least one to-be-audited data corresponding to the data processing rule from the to-be-audited data set according to the data processing rule; and to obtain at least one standard data corresponding to the data processing rule from the standard data set according to the data processing rule; The similarity module is used to obtain the similarity between the data to be reviewed corresponding to the data processing rules and the standard data corresponding to the data processing rules, and determine the review result of the data set to be reviewed corresponding to the data processing rules based on the similarity between the data to be reviewed corresponding to the data processing rules and the standard data corresponding to the data processing rules.

13. The device according to claim 10, wherein: The recall module is used to: Use the 25th algorithm iteration of the best matching algorithm to obtain at least one data to be reviewed from the data set to be reviewed according to at least one preset data processing rule; use the 25th algorithm iteration of the best matching algorithm to obtain at least one standard data from the standard data set according to at least one of the data processing rules.

14. The device according to claim 10, wherein: The similarity module is used for: The similarity between the data to be reviewed and the standard data is obtained by using deep contrastive learning based on unsupervised contrast sentence vector representation.

15. The device according to claim 10, wherein: The data processing device further comprises: The first parsing module is used to perform format parsing and layout parsing on the document to be parsed and reviewed, and obtain the data set to be reviewed; The second parsing module is used to perform format parsing and layout parsing on the standard document to be parsed, and obtain a standard data set.

16. The device according to claim 15, wherein: The data processing device further comprises: The index module is used to insert the to-be-audited data set and the standard data set into corresponding indexes, and save document properties of the documents to which the to-be-audited data set and the standard data set belong.

17. The device according to claim 10, wherein: The data set to be reviewed includes the internal systems formulated by the institution; the standard data set includes the laws and regulations that the institution uses as reference in formulating its internal systems.

18. An electronic device, comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 9.

19. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-9.

20. A computer program product comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 9.

Images