Banking business document consistency checking method and device, equipment and medium

Abstract

The embodiment of the invention discloses a consistency checking method and device for banking business documents, equipment and a medium, and relates to the technical field of document consistency checking. The method comprises the following steps: acquiring a first document and a second document for the same banking business; determining a text consistency check result between the first document and the second document according to the first text vector of the first document and the second text vector of the second document; determining a structural consistency check result between the first document and the second document according to the first graph model of the first document and the second graph model of the second document; and determining a consistency check result of the first document and the second document according to the text consistency check result and the structure consistency check result. According to the technical scheme, consistency judgment of two dimensions of text content and structure is performed on the document through the text vector and the graph model, so that the checking efficiency and the accuracy of a checking result are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of document consistency checking, and in particular to a consistency checking method, device, equipment and medium for banking business documents. Background Art

[0002] In banking, document consistency checking is a key step, which aims to ensure that the document content remains consistent across multiple versions or departments.

[0003] Currently, consistency checks for banking documents typically rely on manual review or simple rule-matching technology. However, manual review is time-consuming and labor-intensive, inefficient, and susceptible to subjective factors. Rule-matching technology, on the other hand, cannot handle complex semantic relationships and structured data, and is prone to misjudgments and omissions. Summary of the Invention

[0004] The present invention provides a consistency check method, apparatus, device and medium for banking business documents, which can perform consistency check on documents based on the text vectors of the documents and a graph model established based on the documents, thereby improving the inspection efficiency and the accuracy of the inspection results.

[0005] According to one aspect of the present invention, a method for checking consistency of a banking document is provided, the method comprising:

[0006] Obtaining a first document and a second document for the same banking transaction;

[0007] determining a text consistency check result between the first document and the second document based on a first text vector of the first document and a second text vector of the second document;

[0008] determining a structural consistency check result between the first document and the second document according to a first graph model of the first document and a second graph model of the second document;

[0009] The consistency check results of the first document and the second document are determined according to the text consistency check result and the structure consistency check result.

[0010] According to another aspect of the present invention, there is provided a device for checking consistency of banking documents, comprising:

[0011] A document acquisition module, configured to acquire a first document and a second document for the same banking business;

[0012] a text consistency check result determination module, configured to determine a text consistency check result between the first document and the second document based on a first text vector of the first document and a second text vector of the second document;

[0013] a structure consistency check result determining module, configured to determine a structure consistency check result between the first document and the second document based on a first graph model of the first document and a second graph model of the second document;

[0014] The consistency check result determination module is configured to determine consistency check results of the first document and the second document according to the text consistency check result and the structure consistency check result.

[0015] According to another aspect of the present invention, an electronic device is provided, comprising:

[0016] at least one processor; and

[0017] a memory communicatively connected to the at least one processor; wherein,

[0018] The memory stores a computer program executable by the at least one processor. The computer program is executed by the at least one processor to enable the at least one processor to perform the consistency checking method for banking business documents according to any embodiment of the present invention.

[0019] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the consistency checking method for banking business documents according to any embodiment of the present invention when executed.

[0020] The technical solution of the embodiment of the present application includes: obtaining a first document and a second document for the same banking business; determining a text consistency check result between the first document and the second document based on a first text vector of the first document and a second text vector of the second document; determining a structural consistency check result between the first document and the second document based on a first graph model of the first document and a second graph model of the second document; and determining a consistency check result between the first document and the second document based on the text consistency check result and the structural consistency check result. This technical solution uses text vectors and graph models to perform consistency judgment on documents in two dimensions: text content and structure, greatly improving inspection efficiency and the accuracy of inspection results.

[0021] 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 invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0023] Figure 1 This is a flowchart of a method for checking consistency of banking business documents provided in accordance with the first embodiment of the present application;

[0024] Figure 2 This is a flow chart of a method for checking consistency of banking business documents provided in accordance with the second embodiment of the present application;

[0025] Figure 3 This is a flowchart of a specific example of consistency check of a banking business document provided in accordance with the second embodiment of the present application;

[0026] Figure 4 This is a schematic diagram of the structure of a consistency check device for banking business documents provided in accordance with the third embodiment of the present application;

[0027] Figure 5 It is a structural diagram of an electronic device that implements a consistency checking method for banking business documents according to an embodiment of the present application. DETAILED DESCRIPTION

[0028] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0029] It should be noted that the terms "first", "second", "target", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0030] Example 1

[0031] Figure 1 A flowchart of a method for checking consistency of banking documents is provided for the first embodiment of the present application. The embodiment of the present application is applicable to situations where consistency checking of banking documents is performed. The method can be performed by a consistency checking device for banking documents. The consistency checking device for banking documents can be implemented in the form of hardware and / or software. The consistency checking device for banking documents can be configured in an electronic device with data processing capabilities. Figure 1 As shown, the method includes:

[0032] S110: Acquire a first document and a second document for the same banking business.

[0033] Among them, the first document and the second document are both documents generated for the same banking business, there is an association between the first document and the second document, there may be differences in the content in the first document and the content in the second document, and the document formats of the first document and the second document may be different, for example, one is a document in PDF format and the other is a document in Word format. For example, in one case, the first document and the second document are both reports under the same banking business, and the difference between the first document and the second document is the version, for example, the second document is a document that has been modified based on the first document. In another case, the first document and the second document may be documents generated by different departments or different personnel for the same banking business. The first document and the second document need to be checked for consistency to determine whether the two documents generated by different departments are consistent, or whether the two documents under different versions are consistent.

[0034] Specifically, in one scenario, obtaining a first document and a second document for the same banking business involves obtaining all documents related to the same banking business, which may be multiple documents, for example, dozens of documents, and identifying any two of them as the first document and the second document. In another scenario, obtaining the latest version and the previous version of a document for the same banking business and identifying these two documents as the first document and the second document, respectively. In yet another scenario, obtaining two documents for the same banking business produced by different departments and identified as the first document and the second document, respectively.

[0035] S120: Determine a text consistency check result between the first document and the second document based on the first text vector of the first document and the second text vector of the second document.

[0036] The first text vector reflects the text content of the first document, and the second text vector reflects the text content of the second document. The text vector may be a vector obtained by vectorizing the document.

[0037] Specifically, after obtaining the first document and the second document, text extraction is performed on the first document and the second document respectively to obtain the extracted text content of the first document and the text content of the second document. In an embodiment of the present application, when performing text extraction on the first document and the second document to obtain the extracted text content of the first document and the text content of the second document, the following steps are included: if the format of the document is a format such as PDF that cannot directly extract text, the document is formatted to obtain a document in a format that can extract text; when extracting the text content, the extracted text content is preprocessed to remove erroneous content, duplicate content, and noise data, etc.

[0038] Furthermore, after obtaining the text content, a text vector can be generated through traditional statistical methods (such as the bag-of-words model); or, a text vector corresponding to the text content can be obtained based on word embedding; or, through deep learning model self-encoding, the hidden state of the last time step of LSTM is used as the text vector.

[0039] The text vector can reflect information about the text content of the document. A similarity calculation is performed on the first text vector and the second text vector, and the obtained similarity calculation result is determined as the text consistency check result.

[0040] S130: Determine a structural consistency check result between the first document and the second document according to the first graph model of the first document and the second graph model of the second document.

[0041] A graph model is a mathematical model that uses a graph structure (nodes and edges) to represent data, relationships, or systems. It abstractly describes the relationships between content within a document through nodes (entities, variables) and edges (relationships or dependencies between nodes). Because graph models reflect the relationships between nodes, they can represent the structural information of banking documents. Therefore, using a document graph model, it is possible to identify structural consistency between two documents.

[0042] Specifically, based on the type of banking business, a graph model that matches the business is determined. This may include standard data models such as Resource Description Frameworks (RDFs) and Property Graphs (LPGs). RDFs is suitable for describing entities / resources, while property graphs are more suitable for solving the problems of efficient querying, using, and storing interconnected data. After determining the graph model framework, nodes and edges are identified in the document, and then a first graph model for the first document and a second graph model for the second document are established. The degree of consistency between the first and second graph models is determined through a graph matching algorithm to obtain a structural consistency check result.

[0043] S140: Determine consistency check results of the first document and the second document according to the text consistency check result and the structure consistency check result.

[0044] Specifically, the text consistency check result reflects the degree of consistency in content between the first document and the second document, and the structural consistency check result reflects the degree of consistency in structure between the first document and the second document. A comprehensive analysis of the two check results can obtain the consistency check results of the first document and the second document.

[0045] In an embodiment of the present application, optionally, the consistency check results of the first document and the second document are determined based on the text consistency check results and the structure consistency check results, including: obtaining a preset text weight and a structure weight; summing the product result of the text weight and the text consistency check result, and the product result of the structure weight and the structure consistency check result to obtain a consistency check result.

[0046] Among them, the text weight and the structure weight are pre-set, and the sum of the text weight and the structure weight can be 1. For example, if the difference between the first document and the second document is that the version number is inconsistent, the text weight is set to be greater than the structure weight, for example, the text weight is 0.9, the structure weight is 0.1, and then the product of the text weight and the text consistency check result, and the product of the structure weight and the structure consistency check result are summed to obtain the consistency check result. This setting is because when only the version number is inconsistent, the difference between the two documents may only be the change of individual texts. In this case, the structures of the two files are almost the same, and the consistency of the text content can be checked in particular.

[0047] The technical solution of the embodiment of the present application includes: obtaining a first document and a second document for the same banking business; determining a text consistency check result between the first document and the second document based on a first text vector of the first document and a second text vector of the second document; determining a structural consistency check result between the first document and the second document based on a first graph model of the first document and a second graph model of the second document; and determining a consistency check result between the first document and the second document based on the text consistency check result and the structural consistency check result. This technical solution uses text vectors and graph models to perform consistency judgment on documents in two dimensions: text content and structure, greatly improving inspection efficiency and the accuracy of inspection results.

[0048] Example 2

[0049] Figure 2 This is a flow chart of a method for checking consistency of banking business documents provided in Example 2 of the present application. This embodiment of the present application is optimized based on the above embodiment.

[0050] like Figure 2 As shown, the method of the embodiment of the present application specifically includes the following steps:

[0051] S210: Acquire a first document and a second document for the same banking business.

[0052] S220: Input the first text vector of the first document and the second text vector of the second document into a text similarity detection model to obtain a text consistency check result.

[0053] In an embodiment of the present application, optionally, the process of determining the text vector of the document includes: cleaning the document to obtain a cleaned document; performing word segmentation on the cleaned document to obtain a segmented text; removing stop words from the segmented text to obtain a text after removing stop words; performing root extraction on the text after removing stop words, and determining the vector of the text after extracting the root words to obtain the text vector of the document.

[0054] This solution is configured such that when generating a text vector for a document, since the document has been pre-processed, such as cleaning, the document content is clearer and more concise, and the extracted text vector can clearly and accurately reflect the document content.

[0055] Exemplarily, after obtaining the first document and the second document, preprocess the documents and extract vectors, which specifically include the following steps: text cleaning, format unification, word segmentation and key information extraction, stop word removal, stemming / lemmatization, and vectorization. When performing text cleaning, noise data in the documents, such as irrelevant characters, duplicate content, special symbols, etc., can be removed to ensure the accuracy of subsequent processing. When performing format unification, convert the documents into a unified format, such as converting PDF to text, removing unnecessary format markers, etc., so that the network model can correctly parse and process the document content. When performing word segmentation and key information extraction, use a word segmentation tool suitable for banking business documents to perform word segmentation on the text. Word segmentation is the first step in text processing, which helps to cut continuous text strings into individual words or phrases for subsequent processing. Use natural language processing technology (NLP) to extract key information in the documents, such as data fields, clause content, signature information, etc., and these information will be used as the basis for subsequent consistency checks. When removing stop words, stop words in the text (such as "de", "le", etc.) can be removed. These words contribute less to text similarity calculation but increase the computational amount. When performing stemming / lemmatization, words can be restored to their basic forms (stems or roots) to reduce the impact of word form changes on similarity calculation. When performing the vectorization step, convert the processed text into a vector representation for similarity calculation. The vectorization methods that can be adopted include but are not limited to: TF-IDF, Word2Vec, BERT, etc.

[0056] In the embodiments of the present application, the text similarity detection model is used to detect the similarity degree of the first text vector and the second text vector, and output the text consistency check result. The determination process of the text similarity detection model includes: determining a basic text similarity model and fine-tuning the basic text similarity model to obtain the text similarity detection model.

[0057] When determining the basic text similarity model, it is necessary to determine the types of documents to be inspected (such as business project applications, financial reports, contracts, etc.), clarify the specific objectives (such as data correctness, clause consistency, format specification, etc.) and specific requirements (such as whether the content is exactly the same, whether the key information is consistent, etc.) of the consistency check. According to banking business specifications and internal requirements, determine detailed consistency check standards, ensure that the standards cover multiple aspects such as document content, format, terms, data, etc., and determine the basic text similarity model accordingly. The basic text similarity model can be a model based on BERT or RoBERTa.

[0058] After determining the basic text similarity model, training data can be obtained. Specifically, this can involve collecting banking business documents, cleaning the documents and standardizing their format to form training and test sets. Documents are then manually annotated to identify which documents have consistent content and which do not.

[0059] Fine-tune the basic text similarity model to obtain a text similarity detection model, including:

[0060] A similarity calculation is performed on each pair of documents that need to be checked for consistency. Appropriate similarity calculation methods can be selected based on business needs, such as cosine similarity, Jaccard similarity coefficient, edit distance, etc. A threshold can be set to determine whether the document content is consistent. If the similarity is above a certain threshold, the content is considered consistent. The trained text similarity detection model is used to perform consistency checks on banking business documents. The check contents include but are not limited to: data consistency: checking whether the data in the document is accurate and consistent, such as amount, date, account number, etc.; clause consistency: checking whether the clauses in the document are consistent with other relevant documents or regulations; format consistency: checking whether the format of the document complies with the specifications; terminology consistency: checking whether the terminology in the document is unified. The network model outputs the check results, including consistency assessment results and potential inconsistencies. The results can be presented in the form of reports, prompts or visualizations.

[0061] Next, evaluate the text similarity detection model to see if it meets business requirements. Use metrics such as the confusion matrix, precision, and recall to assess model performance. Adjust model parameters or retrain the model based on the evaluation results. Conduct in-depth analysis of documents with misclassifications to identify the causes and improve the model. Throughout subsequent use, the model can be regularly updated and optimized to adapt to changes and developments in banking business.

[0062] In an embodiment of the present application, optionally, the first text vector of the first document and the second text vector of the second document are input into a text similarity detection model to obtain a text consistency check result, including: inputting the first text vector of the first document and the second text vector of the second document into a data detection sub-model of the text similarity detection model to obtain a data consistency check result; inputting the first text vector of the first document and the second text vector of the second document into a clause detection sub-model of the text similarity detection model to obtain a clause consistency check result; inputting the first text vector of the first document and the second text vector of the second document into a format detection sub-model of the text similarity detection model to obtain a format consistency check result; inputting the first text vector of the first document and the second text vector of the second document into a term detection sub-model of the text similarity detection model to obtain a term consistency check result; and determining the text consistency check result based on the data consistency check result, the clause consistency check result, the format consistency check result, and the term consistency check result.

[0063] In an embodiment of the present application, the text similarity detection model includes four sub-models, namely: a data detection sub-model, a clause detection sub-model, a format detection sub-model and a term detection sub-model. Specifically, the first text vector of the first document and the second text vector of the second document are input into the four sub-models, and the output results of the four sub-models are obtained respectively. The output results of the four sub-models are summarized to obtain the text consistency check result. This scheme is set up in this way, so that the text similarity detection model can perform consistency checks on the first text vector and the second text vector from the four dimensions of data content, clauses, format and terminology, so that the check results are more accurate, and since four sub-models are set up, each sub-model can focus on the features that it needs to identify, and the recognition ability of each sub-model is stronger.

[0064] S230: Determine a structural consistency check result between the first document and the second document according to the first graph model of the first document and the second graph model of the second document.

[0065] In an embodiment of the present application, optionally, the process of establishing the graph model of the document includes: obtaining the business type of the document; determining the graph model type that matches the business type; and constructing the graph model based on the nodes and edges extracted from the document based on the graph model type.

[0066] Specifically, the banking business type to be inspected is determined, such as loan contracts, financial statements, and business reports. The inspection objectives for documents within this banking business type are determined. For example, these objectives may include: establishing reasonable document structure standards, including the sections a document should contain (such as a cover, table of contents, main text, and attachments), the order in which each section should be arranged, and the logical relationships and hierarchical structure within each section. Furthermore, based on the inspection objectives, a graph model type that matches the business type is determined. Based on this graph model type, a graph model is constructed using the nodes and edges extracted from the document.

[0067] Furthermore, based on the graph model type, a graph model is constructed according to the nodes and edges extracted from the document, including: defining different parts of the document as nodes, and defining the logical relationships between the parts as edges. For example, the cover node can be connected to the directory node, and the directory node is respectively connected to each chapter node, etc. Necessary attributes are added to the nodes and edges in the graph to describe the detailed information of the document structure. For example, a node can contain attributes such as name, type, and location; an edge can contain attributes such as direction and relationship type. Natural language processing (NLP) technology or a specialized document parsing tool is used to parse banking business documents and extract the structural information of the document. When performing document structure parsing, the diversity and complexity of the document format are fully considered, and appropriate parsing tools and methods are selected. The document structure information obtained by parsing is mapped to the graph model corresponding to the graph model type, the different parts of the document are matched with the nodes in the graph model, and corresponding edge and attribute relationships are established to construct a graph model.

[0068] This arrangement of the present solution can construct a graph model that can accurately reflect the structural information in the document. The structural information can be the association relationship between nodes reflected by nodes and edges.

[0069] In an embodiment of the present application, optionally, a structural consistency check result between the first document and the second document is determined based on the first graph model of the first document and the second graph model of the second document, including: calculating the node overlap rate and the edge overlap rate of the first graph model and the second graph model; and determining the product of the node overlap rate and the edge overlap rate as the structural consistency check result between the first document and the second document.

[0070] Specifically, the node overlap ratio reflects the degree of overlap between the first and second graph models at the nodes, while the edge overlap ratio reflects the degree of overlap between the first and second graph models at the edges. The node overlap ratio and the edge overlap ratio can be combined to obtain the structural consistency check result. This arrangement provides a calculation method that quantifies the structural consistency check results and can quickly calculate the structural consistency check results.

[0071] Exemplarily, determining a structural consistency check result between the first document and the second document based on the first graph model of the first document and the second graph model of the second document includes: checking the first graph model of the first document and the second graph model of the second document through a graph matching algorithm, and determining the check result as the structural consistency check result between the first document and the second document.

[0072] The method further includes: adjusting the structural consistency check results into a report format, including a list of checked documents, consistency assessment results for each document, potential inconsistencies and their specific locations, etc. Detailed feedback information is provided for each inconsistency, including the specific content of the inconsistency, possible causes (such as data errors, inconsistent specifications, etc.), and suggested solutions (such as data correction, specification unification), etc. An appropriate report format can be selected based on user needs and preferences, such as a Word document, PDF file, or web page report, etc. Visual tools such as charts and tables are appropriately used in the report to display key information and inconsistencies so that users can quickly understand the problem.

[0073] S240: Determine consistency check results of the first document and the second document according to the text consistency check result and the structure consistency check result.

[0074] Figure 3 The flowchart of a specific example of consistency check of a banking business document includes the following steps:

[0075] 1) Input banking business documents A and B to be checked.

[0076] 2) Use the text similarity model to calculate the similarity between document A and document B to check the content consistency.

[0077] 3) Build a graph model of document A and document B and extract key entities and relationships.

[0078] 4) Apply graph matching algorithm to compare the graph models of two documents and check the structural consistency.

[0079] 5) Combining the results of the text similarity model and the graph model, a consistency check report for document A and document B is obtained.

[0080] The technical solution of the embodiments of this application improves the model's ability to handle complex semantics by establishing a text similarity detection model to check document content consistency. The consistency of the document structure is checked using a graph model, and the document consistency check result is determined based on the document content consistency results and the document structure consistency results. This allows for a comprehensive check of the document, with a high degree of automation and high accuracy of the check results.

[0081] Example 3

[0082] Figure 4 This is a schematic diagram of the structure of a consistency check device for a banking business document provided in the third embodiment of the present application. The device can execute the consistency check method for banking business documents provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method. Figure 4 As shown, the device includes:

[0083] A document acquisition module 310 is configured to acquire a first document and a second document for the same banking business;

[0084] a text consistency check result determination module 320, configured to determine a text consistency check result between the first document and the second document based on the first text vector of the first document and the second text vector of the second document;

[0085] a structure consistency check result determining module 330, configured to determine a structure consistency check result between the first document and the second document based on a first graph model of the first document and a second graph model of the second document;

[0086] The consistency check result determination module 340 is configured to determine consistency check results of the first document and the second document according to the text consistency check result and the structure consistency check result.

[0087] The technical solution of the embodiment of the present application includes: a document acquisition module 310, which is used to acquire a first document and a second document for the same banking business; a text consistency check result determination module 320, which is used to determine the text consistency check result between the first document and the second document based on the first text vector of the first document and the second text vector of the second document; a structural consistency check result determination module 330, which is used to determine the structural consistency check result between the first document and the second document based on the first graph model of the first document and the second graph model of the second document; a consistency check result determination module 340, which is used to determine the consistency check result of the first document and the second document based on the text consistency check result and the structural consistency check result. This technical solution uses text vectors and graph models to perform consistency judgment on documents in two dimensions: text content and structure, which greatly improves the inspection efficiency and the accuracy of the inspection results.

[0088] Optionally, the consistency check result determination module 340 includes:

[0089] A weight acquisition unit, used to acquire preset text weights and structure weights;

[0090] The consistency check result determining unit is configured to sum the product of the text weight and the text consistency check result and the product of the structure weight and the structure consistency check result to obtain a consistency check result.

[0091] Optionally, the device further includes: a text vector determination module, comprising:

[0092] A document cleaning unit, used for cleaning the document to obtain a cleaned document;

[0093] The document segmentation unit is used to perform word segmentation on the cleaned document to obtain the segmented text;

[0094] A stop word removal unit is used to remove stop words from the text after word segmentation to obtain a text after the stop words are removed;

[0095] The root extraction unit is used to extract the root of the text after removing the stop words, and determine the vector of the text after the root is extracted to obtain the text vector of the document.

[0096] In the embodiment of the present application, optionally, the text consistency check result determination module 320 includes:

[0097] The text consistency check result determining unit is used to input the first text vector of the first document and the second text vector of the second document into a text similarity detection model to obtain a text consistency check result.

[0098] In an embodiment of the present application, optionally, the text consistency check result determination unit includes:

[0099] a data detection subunit, configured to input the first text vector of the first document and the second text vector of the second document into a data detection submodel of a text similarity detection model to obtain a data consistency check result;

[0100] a clause detection subunit, configured to input the first text vector of the first document and the second text vector of the second document into a clause detection submodel of a text similarity detection model to obtain a clause consistency check result;

[0101] a format detection subunit, configured to input the first text vector of the first document and the second text vector of the second document into a format detection submodel of a text similarity detection model to obtain a format consistency check result;

[0102] a term detection subunit, configured to input the first text vector of the first document and the second text vector of the second document into a term detection submodel of a text similarity detection model to obtain a term consistency check result;

[0103] The text consistency check result determination subunit is used to determine the text consistency check result according to the data consistency check result, the clause consistency check result, the format consistency check result and the terminology consistency check result.

[0104] Optionally, the structure consistency check result determination module 330 includes:

[0105] An overlap rate calculation unit, configured to calculate a node overlap rate and an edge overlap rate of the first graph model and the second graph model;

[0106] The structural consistency check result determining unit is configured to determine the product of the node overlap rate and the edge overlap rate as the structural consistency check result between the first document and the second document.

[0107] In the embodiment of the present application, optionally, the apparatus further includes: a graph model establishment module, including:

[0108] A business type acquisition unit, used to acquire the business type of a document;

[0109] A graph model type determining unit, configured to determine a graph model type that matches the business type according to the business type;

[0110] The graph model construction unit is configured to construct a graph model based on the graph model type and the nodes and edges extracted from the document.

[0111] The consistency checking device for a banking business document provided in an embodiment of the present application can execute the consistency checking method for a banking business document provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0112] Example 4

[0113] Figure 5 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention 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 (such as helmets, glasses, watches, etc.) 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 invention described and / or claimed herein.

[0114] like Figure 5As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., which is communicatively connected to the at least one processor 11. The memory stores a computer program that can be executed by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. Various programs and data required for the operation of the electronic device 10 can also be stored in the RAM 13. The processor 11, ROM 12, and RAM 13 are connected to each other via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0115] Multiple components in the electronic device 10 are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0116] Processor 11 can be any general-purpose and / or specialized processing component with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any other suitable processor, controller, microcontroller, etc. Processor 11 executes the various methods and processes described above, such as the consistency check method for banking business documents.

[0117] In some embodiments, the consistency checking method for banking documents can be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the consistency checking method for banking documents described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform the consistency checking method for banking documents via any other suitable means (e.g., via firmware).

[0118] Various embodiments of the systems and techniques described above 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 a chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes 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.

[0119] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer program may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0120] In the context of the present invention, computer-readable storage media can be tangible media that can contain or store a computer program for use with an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. Computer-readable storage media can include but are not limited to electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, computer-readable storage media can be machine-readable signal media. More specific examples of machine-readable storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0121] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device 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 electronic device. 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).

[0122] The systems and techniques described herein can 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 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 can 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), a blockchain network, and the Internet.

[0123] A computing system may include clients and servers. The clients and servers are typically remote from each other and typically interact via a communication network. This client-server relationship arises through computer programs running on the respective computers, creating a client-server relationship. The server may be a cloud server, also known as a cloud computing server or cloud host. This server is a hosting product within a cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosting and VPS services.

[0124] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the present invention can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved. This is not limited herein.

[0125] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A method for checking consistency of banking business documents, characterized in that: include: Obtaining a first document and a second document for the same banking transaction; determining a text consistency check result between the first document and the second document based on a first text vector of the first document and a second text vector of the second document; determining a structural consistency check result between the first document and the second document according to a first graph model of the first document and a second graph model of the second document; The consistency check results of the first document and the second document are determined according to the text consistency check result and the structure consistency check result.

2. The method according to claim 1, characterized in that Determining consistency check results of the first document and the second document according to the text consistency check result and the structure consistency check result includes: Get the preset text weight and structure weight; The product of the text weight and the text consistency check result and the product of the structure weight and the structure consistency check result are summed to obtain a consistency check result.

3. The method according to claim 1, characterized in that The process of determining the text vector of the document includes: Cleaning the document to obtain a cleaned document; Perform word segmentation on the cleaned document to obtain the segmented text; Remove stop words from the text after word segmentation to obtain the text after removing stop words; Perform word root extraction on the text after removing stop words, determine the vector of the text after word root extraction, and obtain the text vector of the document.

4. The method according to claim 1, wherein Determining a text consistency check result between the first document and the second document according to a first text vector of the first document and a second text vector of the second document includes: The first text vector of the first document and the second text vector of the second document are input into a text similarity detection model to obtain a text consistency check result.

5. The method according to claim 4, characterized in that Inputting the first text vector of the first document and the second text vector of the second document into a text similarity detection model to obtain a text consistency check result, including: Inputting the first text vector of the first document and the second text vector of the second document into a data detection sub-model of a text similarity detection model to obtain a data consistency check result; Inputting the first text vector of the first document and the second text vector of the second document into a clause detection sub-model of a text similarity detection model to obtain a clause consistency check result; Inputting the first text vector of the first document and the second text vector of the second document into a format detection submodel of a text similarity detection model to obtain a format consistency check result; Inputting the first text vector of the first document and the second text vector of the second document into a term detection submodel of a text similarity detection model to obtain a term consistency check result; A text consistency check result is determined based on the data consistency check result, the clause consistency check result, the format consistency check result, and the terminology consistency check result.

6. The method according to claim 1, characterized in that Determining a structural consistency check result between the first document and the second document according to a first graph model of the first document and a second graph model of the second document includes: Calculating the node overlap rate and the edge overlap rate of the first graph model and the second graph model; The product of the node overlap rate and the edge overlap rate is determined as a structural consistency check result between the first document and the second document.

7. The method according to claim 1, characterized in that The process of establishing the document graph model includes: Get the business type of the document; According to the business type, determining a graph model type that matches it; Based on the graph model type, a graph model is constructed according to the nodes and edges extracted from the document.

8. A consistency check device for banking business documents, characterized in that: include: A document acquisition module, configured to acquire a first document and a second document for the same banking business; a text consistency check result determination module, configured to determine a text consistency check result between the first document and the second document based on a first text vector of the first document and a second text vector of the second document; a structure consistency check result determining module, configured to determine a structure consistency check result between the first document and the second document based on a first graph model of the first document and a second graph model of the second document; The consistency check result determination module is configured to determine consistency check results of the first document and the second document according to the text consistency check result and the structure consistency check result.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor. The computer program is executed by the at least one processor to enable the at least one processor to perform the consistency checking method for banking business documents according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the consistency checking method for banking business documents according to any one of claims 1 to 7 when executed.

Images