A method and system for examining documents under a letter of credit

Abstract

The present invention provides a method and system for examining documents under a letter of credit, which acquire documents to be examined and a letter of credit to be examined; based on a pre-constructed document examination knowledge graph, extract document entity attribute structure data from the documents to be examined, and extract clause intentions and slot data from the letter of credit to be examined; input the document entity attribute structure data, clause intentions and slot data into a document examination execution engine, so that the document examination execution engine examines the document entity attribute structure data, clause intentions and slot data based on document examination rules to obtain an examination result of the documents to be examined and the letter of credit to be examined. In this solution, by using the document examination knowledge graph, document entity attribute structure data is extracted from the documents to be examined, and clause intentions and slot data are extracted from the letter of credit to be examined. The document entity attribute structure data, clause intentions and slot data are input into the document examination execution engine for examination to obtain an examination result, without the need for manual examination of the letter of credit and documents, improving the examination efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to a method and system for examining and verifying letters of credit Background Art

[0002] Cross-border trade is a traditional type of trade, and the letter of credit business has been fully integrated into cross-border trade operations. To better conduct cross-border trade operations, it is necessary to implement the examination and verification of the consistency between letters of credit and documents.

[0003] The existing methods for examining and verifying documents mainly rely on manual review. However, manual review requires a large amount of time and is prone to errors, resulting in poor review efficiency. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a method and system for examining and verifying letters of credit to solve problems such as poor review efficiency existing in the existing document examination and verification methods.

[0005] To achieve the above object, embodiments of the present invention provide the following technical solutions:

[0006] In a first aspect, an embodiment of the present invention discloses a method for examining and verifying letters of credit, the method comprising:

[0007] Obtain the document to be examined and verified and the letter of credit to be examined and verified;

[0008] Based on a pre-constructed knowledge graph for document examination and verification, extract the document entity attribute structure data from the document to be examined and verified, and extract the clause intention and slot data from the letter of credit to be examined and verified. The knowledge graph for document examination and verification at least includes: static knowledge of letters of credit and documents, document examination rules, and a knowledge base for letter of credit document examination;

[0009] Input the document entity attribute structure data, the clause intention, and the slot data into a document examination execution engine, and enable the document examination execution engine to examine and verify the document entity attribute structure data, the clause intention, and the slot data based on the document examination rules to obtain the examination and verification results of the document to be examined and verified and the letter of credit to be examined and verified.

[0010] Preferably, the process of extracting the document entity attribute structure data from the document to be examined and verified includes:

[0011] Based on the static knowledge of letters of credit and documents and the knowledge base for letter of credit document examination in the pre-constructed knowledge graph for document examination and verification, extract the document entity attribute structure data from the document to be examined and verified through an entity protocol converter.

[0012] Preferably, extracting the clause intention and slot data from the letter of credit to be examined and verified includes:

[0013] Based on the static knowledge of letters of credit and documents and the letter of credit examination knowledge base in the pre-constructed examination knowledge graph, use a semantic analysis engine to extract clause intentions and slot data from the letter of credit to be examined.

[0014] Preferably, the method further includes:

[0015] When detecting a first operation on the examination rules of the examination knowledge graph, execute the first operation to add, delete, or modify examination rules.

[0016] Preferably, the method further includes:

[0017] When detecting a second operation on the static knowledge of letters of credit and documents in the examination knowledge graph, execute the second operation to define, query, modify, or delete the content in the static knowledge of letters of credit and documents.

[0018] Preferably, the method further includes:

[0019] When detecting a third operation on the letter of credit examination knowledge base of the examination knowledge graph, execute the third operation to modify or add the content in the letter of credit examination knowledge base.

[0020] A second aspect of the embodiments of the present invention discloses a letter of credit examination system, the system includes:

[0021] An acquisition unit, configured to acquire documents to be examined and letters of credit to be examined;

[0022] An extraction unit, configured to extract document entity attribute structure data from the documents to be examined and extract clause intentions and slot data from the letters of credit to be examined based on a pre-constructed examination knowledge graph, where the examination knowledge graph at least includes: static knowledge of letters of credit and documents, examination rules, and a letter of credit examination knowledge base;

[0023] An examination unit, configured to input the document entity attribute structure data, the clause intentions, and the slot data into an examination execution engine, so that the examination execution engine examines the document entity attribute structure data, the clause intentions, and the slot data based on the examination rules to obtain the examination results of the documents to be examined and the letters of credit to be examined.

[0024] Preferably, the extraction unit for extracting document entity attribute structure data from the documents to be examined is specifically configured to: based on the static knowledge of letters of credit and documents and the letter of credit examination knowledge base in the pre-constructed examination knowledge graph, extract document entity attribute structure data from the documents to be examined through an entity protocol converter.

[0025] Preferably, the extraction unit for extracting clause intents and slot data from the letter of credit to be audited is specifically configured to: based on the static knowledge of letters of credit and documents and the knowledge base for letter of credit document review in the pre-constructed document review knowledge graph, use a semantic analysis engine to extract clause intents and slot data from the letter of credit to be audited.

[0026] Preferably, the system further includes:

[0027] A first execution unit, configured to execute the first operation to add, delete, or modify a document review rule when detecting a first operation on the document review rule for the document review knowledge graph.

[0028] Based on the letter of credit document review method and system provided in the above embodiments of the present invention, the method is as follows: obtain a document to be audited and a letter of credit to be audited; based on the pre-constructed document review knowledge graph, extract document entity attribute structure data from the document to be audited, and extract clause intents and slot data from the letter of credit to be audited; input the document entity attribute structure data, clause intents, and slot data into a document review execution engine, so that the document review execution engine reviews the document entity attribute structure data, clause intents, and slot data based on the document review rules to obtain the review results of the document to be audited and the letter of credit to be audited. In this solution, by using the document review knowledge graph, document entity attribute structure data is extracted from the document to be audited, and clause intents and slot data are extracted from the letter of credit to be audited. The document entity attribute structure data, clause intents, and slot data are input into the document review execution engine for review to obtain the review results, eliminating the need for manual review of letters of credit and documents and improving the review efficiency. Description of the Drawings

[0029] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on the provided drawings without creative efforts.

[0030] Figure 1 It is a flowchart of a letter of credit document review method provided by an embodiment of the present invention;

[0031] Figure 2 It is a schematic flowchart of extracting document entity attribute structure data provided by an embodiment of the present invention;

[0032] Figure 3 It is a flow example diagram of a letter of credit document review method provided by an embodiment of the present invention;

[0033] Figure 4It is an example diagram of the concepts included in the static knowledge of letters of credit and documents provided by the embodiments of the present invention;

[0034] Figure 5(a) is an example diagram of the letter of credit data baseline provided by the embodiments of the present invention; Figure 5(b) is an example diagram of the invoice data baseline provided by the embodiments of the present invention; Figure 5(c) is an example diagram of the bill of lading data baseline provided by the embodiments of the present invention; Figure 5(d) is an example diagram of the CCL provided by the embodiments of the present invention; Figure 5(e) is an example diagram of the business attributes provided by the embodiments of the present invention;

[0035] Figure 6 It is a flowchart for determining the document examination rules provided by the embodiments of the present invention;

[0036] Figure 7 It is a schematic diagram of the business architecture of the letter of credit document examination method provided by the embodiments of the present invention;

[0037] Figure 8 It is a hierarchical diagram of the technical architecture of the letter of credit document examination method provided by the embodiments of the present invention;

[0038] Figure 9 It is a structural block diagram of a letter of credit document examination system provided by the embodiments of the present invention. Detailed implementation manners

[0039] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0040] In this application, the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, the element defined by the statement "including a..." does not exclude the existence of another identical element in the process, method, article or device including the said element.

[0041] As can be seen from the background technology, the current document examination method mainly relies on manual examination, which consumes a lot of time and is prone to mistakes during document examination, and the examination efficiency is poor.

[0042] Therefore, an embodiment of the present invention provides a method and system for examining documents of a letter of credit. By using a knowledge graph for document examination, it extracts data on the entity attribute structure of documents from the documents to be examined, and extracts clause intentions and slot data from the letter of credit to be examined. The data on the entity attribute structure of documents, clause intentions, and slot data are input into a document examination execution engine for examination to obtain an examination result, eliminating the need for manual examination of the letter of credit and documents, thereby improving the examination efficiency.

[0043] See Figure 1 , which shows a flowchart of a method for examining documents of a letter of credit provided by an embodiment of the present invention. The method for examining documents of a letter of credit includes:

[0044] Step S101: Obtain the documents to be examined and the letter of credit to be examined.

[0045] In the specific implementation of step S101, obtain the documents to be examined and the letter of credit to be examined that need to be examined.

[0046] Step S102: Based on a pre-constructed knowledge graph for document examination, extract data on the entity attribute structure of documents from the documents to be examined, and extract clause intentions and slot data from the letter of credit to be examined.

[0047] It should be noted that based on cognitive linguistics, logic, terminology, standardization knowledge, and the basic principles of knowledge graphs, etc., a knowledge graph for document examination (which can express document business knowledge) is constructed from document examination knowledge such as document examination practice, UCP600, ISBP745, and International Chamber of Commerce cases.

[0048] The knowledge graph for document examination at least includes: static knowledge of letters of credit and documents, document examination rules, and a knowledge base for examining letters of credit; or rather, the knowledge graph for document examination consists of these three parts: static knowledge of letters of credit and documents, document examination rules (also known as dynamic knowledge of document examination rules or examination rules), and a knowledge base for examining letters of credit. Through the knowledge graph for document examination, the relationships between documents, document sets (documents and letters of credit), and document pairs (documents and documents) can be provided.

[0049] Furthermore, it should be noted that the knowledge graph for document examination defines which data need to be extracted for document examination (which can be regarded as a complete set here); in the specific implementation of step S102, based on the knowledge graph for document examination, extract data on the entity attribute structure of documents (equivalent to a subset of the aforementioned complete set) from the documents to be examined, and extract clause intentions (equivalent to a subset of the aforementioned complete set) and slot data (equivalent to a subset of the aforementioned complete set) from the letter of credit to be examined.

[0050] In some embodiments, based on the static knowledge of letters of credit and documents and the knowledge base for examining letters of credit in the pre-constructed knowledge graph for document examination, data on the entity attribute structure of documents are extracted from the documents to be examined through an entity protocol converter. Specifically, such asFigure 2 The flowchart shows the process of extracting the structured data of the document entity attributes. Steps such as document parsing, document text information extraction, and substitution of document element values are performed on the document to be audited to obtain the structured data of the document entity attributes (also known as the knowledge-based representation of the document). Among them, when performing the step of document text information extraction, the content of the knowledge model library, domain knowledge representation, and general knowledge representation needs to be called. Through the above method, the structured document can be converted into a unified knowledge-based document model representation for subsequent business processing.

[0051] In some other embodiments, based on the static knowledge of letters of credit and documents in the pre-constructed document audit knowledge graph and the letter of credit audit knowledge base, a semantic analysis engine is used to extract the clause intent and slot data from the letter of credit to be audited.

[0052] Step S103: Input the structured data of the document entity attributes, clause intent, and slot data into the document audit execution engine, so that the document audit execution engine audits the structured data of the document entity attributes, clause intent, and slot data based on the audit rules to obtain the audit results of the document to be audited and the letter of credit to be audited.

[0053] In some embodiments, it is determined that the document audit knowledge graph is the knowledge constraint framework of the document audit execution engine; at least the audit rules of the document audit knowledge graph are set in the document audit execution engine.

[0054] In the process of specifically implementing step S103, the structured data of the document entity attributes, clause intent, and slot data are input into the document audit execution engine; the document audit execution engine audits the structured data of the document entity attributes, clause intent, and slot data based on the audit rules to obtain the audit results of the document to be audited and the letter of credit to be audited, and the audit results include one or more of the compliance points, non-compliance points, and prompt points.

[0055] For a better explanation of the content of steps S101 to S103, a flowchart example of a letter of credit audit method is shown through Figure 3 ; as Figure 3 , the structured data of the document and the letter of credit terms in the letter of credit business instance correspond to the document to be audited and the letter of credit to be audited respectively; the structured data of the document entity attributes are extracted from the document to be audited through the entity protocol converter, and the clause intent and slot data are extracted from the letter of credit to be audited by using the semantic analysis engine. The pre-constructed and published document audit knowledge graph is the knowledge constraint framework of the document audit execution engine, and there are multiple document audit engine components in the document audit execution engine. Input the structured data of the document entity attributes, clause intent, and slot data into the document audit execution engine for auditing, and the audit results (that is, the document audit results) can be output.

[0056] In an embodiment of the present invention, by using the bill review knowledge graph, the bill entity attribute structure data is extracted from the bill to be reviewed, and the clause intention and slot data are extracted from the letter of credit to be reviewed. The bill entity attribute structure data, the clause intention and the slot data are input into the bill review execution engine for review to obtain a review result, without the need for manual review of the letter of credit and the bill, thus improving the review efficiency.

[0057] For the letter of credit and bill static knowledge, bill review rules and letter of credit review knowledge base included in the above-mentioned bill review knowledge graph, the following content will be used to explain and illustrate them respectively.

[0058] Explanation of the letter of credit and bill static knowledge:

[0059] It should be noted that the letter of credit and bill static knowledge defines the concepts involved in the international settlement bill review business process and the relationships between them. The core business object of international settlement bill review is the letter of credit transaction. The letter of credit transaction covers the letter of credit, the bills under the letter of credit transaction, and the participants in the transaction, etc.; from a business perspective, Figure 4 , the letter of credit and bill static knowledge mainly covers the following levels of concepts and relationships: "classification and subdivision of attribute values", "bill attributes and attribute values, attributes and attribute values of participants", "letter of credit transaction (concepts involved in bills and participants)".

[0060] In the above concepts, the letter of credit, bills (such as invoices and bills of exchange, etc.), and transaction participants (such as trading companies, banks, and insurance companies, etc.) are usually regarded as classes; the elements and features shown on the letter of credit and bills are regarded as the attributes of the corresponding classes. It should be noted that there is a certain upper and lower relationship between classes, and the lower-level concepts can inherit the attributes of the upper-level concepts. For example: an invoice is a type of bill, so the bill is the upper term and the invoice is the lower term. The bill has attribute values such as issue date and issuer, then the invoice inherits the attributes of the bill, that is, the invoice also has attribute values such as issue date and issuer.

[0061] An attribute represents a special relationship, usually used to characterize the features of a certain concept. Analyzing the attributes of letters of credit, documents, and parties can further divide sub-attributes (if necessary) or abstract common attributes; there are two types of attribute values pointed to by attributes, one is a value and the other is an entity; "value" is used to describe the features of an entity / concept, for example: the number of a document; "entity" is used to describe the relationship between an entity / concept and other entities / concepts, for example: the applicant of a letter of credit is a trading company entity, which has attributes such as name and address; for attributes with "entity" as the value, entity classes can be refined from these entities, such as continents, countries, cities, seaports, airports, standard measurement units, and commodities; for attributes with "value" as the value, the "value" classes can be analyzed, for example: letter of credit number and issuing date, and analyze the value range and legality requirements of the letter of credit number and issuing date; for values of the enumeration type, find the value range set.

[0062] It should be noted that the ontology specifically refers to the concept layer of the bill review knowledge graph; the SCHEMA layer of the ontology is a concept framework; the static knowledge of letters of credit and documents is the representation of letters of credit under the ontology framework.

[0063] The SCHEMA layer of the ontology is designed based on the "metadata" of ISO7372 as the data baseline combined with the business data of letter of credit bill review practice to establish prototype templates for the letter of credit data baseline, invoice data baseline, and bill of lading data baseline; the "baseline" mentioned above is a business data representation designed by the ISO organization.

[0064] For example: The example content shown in Figure 5(a) is part of the relevant content of the letter of credit data baseline; the example content shown in Figure 5(b) is part of the relevant content of the invoice data baseline; the example content shown in Figure 5(c) is part of the relevant content of the bill of lading data baseline.

[0065] It can be understood that the hierarchical structure of attributes describing business data is designed based on the framework of the Core Component Library (CCL) combined with the practice of letter of credit business. For example, the example diagram of CCL provided in Figure 5(d) shows some example content of CCL representation.

[0066] Business attributes specify the format of data. For example: The AMOUNT class contains currency and value. The letter of credit amount of a letter of credit belongs to the AMOUNT type, so the currency and value need to be parsed during letter of credit parsing. All concepts in the static knowledge of letters of credit and documents define business attributes, which are used to represent a unified data parsing structure for review; for example, the example diagram of business attributes provided in Figure 5(e) shows some example content of business attribute representation.

[0067] Static knowledge of letters of credit and documents includes: knowledge of letter of credit elements, knowledge of bill of exchange elements, knowledge of invoice elements, knowledge of bill of lading elements, knowledge of insurance policy elements, knowledge of packing list elements, knowledge of weight list elements, knowledge of empty list elements, knowledge of charter party bill of lading elements, knowledge of multimodal transport bill of lading elements, knowledge of sea waybill elements, knowledge of beneficiary's certificate elements, knowledge of certificate of origin elements, knowledge of shipping company's certificate elements, knowledge of shipping certificate elements, knowledge of analysis certificate elements, knowledge of quality certificate elements, knowledge of inspection certificate elements, knowledge of quantity certificate elements, knowledge of phytosanitary certificate elements, knowledge of health certificate elements, knowledge of packing certificate elements, knowledge of fumigation certificate elements, knowledge of express certificate elements, knowledge of cargo receipt elements, knowledge of freight forwarder's cargo receipt elements, a total of 26 categories. The granularity of the knowledge representation of the above 26 types of documents is consistent with the granularity of document examination. The content to be represented not only includes the concepts pointed to by the documents and the elements on the documents, the language-level knowledge corresponding to the concepts, and the attribute relationships between the concepts, but also includes the verbs, conditions, and modal logics reflected in the documents and the letter of credit terms. The elements that need to be represented in the static knowledge of letters of credit and documents altogether depict an ontology framework containing more than 200 basic types, more than 700 attributes, and more than 200 statement classes (that is, the scope of the static knowledge of letters of credit and documents), forming more than 80,000 knowledge entities. The quantities of the above basic types, attributes, and statement classes are only for illustration.

[0068] The above content is the relevant description of the static knowledge of letters of credit and documents.

[0069] Explanation of document examination rules (or dynamic knowledge of document examination rules):

[0070] The document examination rule representation framework supports the description and definition of examination items and each examination rule under the examination item. The representation of document examination rules needs to follow certain grammar specifications. To ensure the reliability of document examination, deductive logic is used for the representation of document examination logic knowledge. The rule is represented as "A->B" based on a formal ontology. Among them, the antecedent A is a logical expression composed of the combination of (operators, ontology) representing the dimensions of the examination scenario, and the consequent B is the examination logic represented by the combination of (operators, ontology). When A is true, it is deduced that the consequent B is executed. The specific examination logic can be represented by the following content.

[0071] If "TRUE->B", it means that the B examination logic will be executed in all scenarios, such as the mandatory item examination within the document.

[0072] If "A -> B", it represents the review of the matching scenario, and each dimension in scenario A is represented by the conjunction of the letter of credit attributes, document attributes, and clause attributes through operators. For example, if the letter of credit clause is "The letter of credit requires the bill of lading to show the letter of credit number", it is converted into a rule as "exist(DocumentaryCredit.clauses.BillLadingRequiredClause.showCreditNumbe r) -> exist(BillLading.creditNumber) && equal(BillLading.creditNumber, Documenta ryCredit.creditNumber)".

[0073] If "A -> FALSE", it represents the matching scenario and directly outputs the discrepant result.

[0074] If "A -> TRUE", it represents the matching scenario and directly outputs the compliant result.

[0075] The dynamic knowledge of document review rules includes the review requirements of 25 types of documents (bill of exchange, invoice, bill of lading, insurance policy, packing list, weight list, air waybill, charter party bill of lading, multimodal transport bill of lading, sea waybill, beneficiary's certificate, certificate of origin, shipping company's certificate, certificate of shipment, analysis certificate, quality certificate, inspection certificate, quantity certificate, phytosanitary certificate, health certificate, packing certificate, fumigation certificate, courier certificate, cargo receipt, forwarder's cargo receipt) such as UCP / ISBP, international practices, business practices, and letters of credit.

[0076] For each type of document, the review requirements of this type of document include: the compliance review rules of the document itself, the "document and credit conforming" review rules consistent with the letter of credit requirements, and the "document and document conforming" review rules between other documents. Divided by document dimension, the content to be represented by the dynamic knowledge of letter of credit review rules includes the review rules of more than 2,900 business scenarios of 25 types of documents.

[0077] It can be understood that the definition of document review rules is carried out under the working template of the review rule representation framework agreed upon in the business, and the business review items are converted into a computer - understandable document review logic processing process based on a custom rule grammar combined with ontology representation; for the specific process of defining document review rules, see Figure 6 the flow chart showing the determination of document review rules Figure 6 including the following steps:

[0078] Step S601: At the business level, sort out the review items according to the rule representation template.

[0079] It should be noted that the rule representation template is used to design document review rules, and the review items are the categories in the document review rules.

[0080] Step S602: Conduct a business review on the reviewed items obtained through sorting. If the business review is passed, execute Step S603; if the business review is not passed, return to execute Step S601.

[0081] In specific implementation, use the bill review execution engine to conduct a business review on the reviewed items obtained through sorting; if the bill review execution engine can output a review result, it means that the business review is passed.

[0082] Step S603: Analyze the review logic of the business description for the reviewed items passed in the review to identify the required concepts and attributes.

[0083] It should be noted that the identified required concepts and attributes are the above-mentioned bill entity attribute structure data, clause intention, and slot data.

[0084] Step S604: Determine whether the ontology meets the preset conditions. If the ontology meets the preset conditions, reuse the ontology concepts and attributes, and execute Step S605; if the ontology does not meet the preset conditions, update the ontology and execute Step S605.

[0085] It should be noted that the precondition is the condition required by the current rule expression.

[0086] Step S605: Identify the precondition of the bill review rule from the review logic of the business description, and describe it based on the rule grammar to form the antecedent of the bill review rule.

[0087] Step S606: Determine whether the grammar of the bill review rule supports scenario expression. If it supports scenario expression, execute Step S607; if it does not support scenario expression, improve the grammar of the bill review rule, and return to execute Step S605.

[0088] It should be noted that common bill review action operators have been predefined in the grammar of the bill review rule. If the operators in the existing grammar do not meet the scenario expression, a grammar improvement requirement (the processing logic and use case description of the operator need to be described) can be submitted to improve the existing grammar.

[0089] Step S607: Identify the processing actions that the bill review rule needs to execute from the review logic of the business description, and express and describe them based on the rule grammar to form the consequent of the bill review rule.

[0090] Step S608: Determine whether the grammar of the bill review rule supports the review action expression. If it supports the review action expression, execute Step S609; if it does not support the review action expression, improve the grammar of the bill review rule, and return to execute Step S607.

[0091] Step S609: Determine the rule representation of the bill review rule.

[0092] Step S610: Use a regular grammar verification tool to verify the rule representation of the document review rule and determine whether the rule representation passes the verification. If it passes the verification, execute Step S611; if it fails the verification, return to Step S609 to determine the rule representation of the document review rule again.

[0093] Specifically, use a regular grammar verification tool to verify the legality of the grammar of the document review rule and the validity of the concepts and attributes in the document review rule.

[0094] Step S611: Assign a rule number to the document review rule and register it in the review rule representation framework document.

[0095] The above content is the relevant description of the document review rule.

[0096] Description of the letter of credit document review knowledge base:

[0097] It should be noted that due to the high complexity and strong professionalism of the international settlement letter of credit document review, in order to ensure the accuracy of the review information, it is necessary to establish a letter of credit document review knowledge base to support intelligent document review.

[0098] The entities in the letter of credit document review knowledge base (which can also be called the basic knowledge base) have the following three specific characteristics.

[0099] The first characteristic: the namable characteristic. Entities such as trading companies, banks, countries, ports, etc. must have a name attribute, and there may be multiple aliases in the real world.

[0100] The second characteristic: the cross-scenario characteristic. The entities in the letter of credit document review knowledge base objectively exist, but they play different roles in different scenarios. For example, a trading company is a buyer or a seller in a transaction scenario, and may be an applicant or a beneficiary in a letter of credit opening scenario.

[0101] The third characteristic: the reusable characteristic. Due to the objectivity of the entities in the letter of credit document review knowledge base, in addition to being used in letter of credit document review applications, they can also be used as basic entities in other fields such as foreign exchange and insurance.

[0102] All kinds of entity dictionaries can be generated through data collation, and the entity dictionaries are converted and stored in the graph system to form a letter of credit document review knowledge base; the type name of the letter of credit document review knowledge base exists in the ontology framework definition.

[0103] Among them, the above-generated entity dictionaries include but are not limited to: continent dictionary, national geography dictionary, administrative division geography dictionary, international seaport entity dictionary, international airport entity dictionary, international highway station entity dictionary, international railway station entity dictionary, international inland port entity dictionary, bank entity dictionary, CCB international settlement 33 institutional tree entity dictionary, trading company entity dictionary, transportation company entity dictionary, insurance company entity dictionary, official and independent certification agency entity dictionary, measurement unit dictionary, insurance type dictionary, commodity name entity dictionary, a total of 17. The letter of credit document review knowledge base is mainly used to assist in supporting entity information understanding, extraction, and entity linking functions.

[0104] The above content is the relevant description of the letter of credit document review knowledge base.

[0105] After constructing the document review knowledge graph, the content of the document review knowledge graph can be supplemented and maintained through the graph management tool. The graph management tool at least supports multi-person collaboration functions and human-computer interaction interface functions. Specifically, in response to the operations triggered by business operators through the graph management tool, the content of the document review knowledge graph is supplemented and maintained.

[0106] In some embodiments, when a first operation on the document review rules of the document review knowledge graph is detected, the first operation is executed to add, delete, or modify the document review rules. In practical applications, business operators operate the graph management tool through a browser to access the document review rules of the document review knowledge graph. Business personnel can trigger the first operation through the graph management tool to add, delete any document review rules, or update any document review rules. At the same time, the graph management tool supports the configuration of rule application conditions to quickly filter the document review rules applicable to the current instance for the document review execution engine. Business operators can also trigger the first operation through the graph management tool to classify and manage the document review rules, such as supporting the classification directory function; each document review rule supports freely adding text descriptions, such as adding traceability descriptions, that is, expressing which clause of the convention the rule is defined according to. When each document review rule is maintained, a conflict check and verification mechanism for the document review rules is required. The maintenance operations of the document review rules are graded according to the risk level and authorized by personnel with different permission roles.

[0107] In some other embodiments, when a second operation on the static knowledge of letters of credit and documents in the bill review knowledge graph is detected, the second operation is executed to define, query, modify, or delete the content in the static knowledge of letters of credit and documents. In practical applications, business operators access the content in the static knowledge of letters of credit and documents through a browser to operate the graph management tool, and can perform maintenance operations such as defining, querying, modifying, and deleting concept classes, attributes, relationships between concepts, constraints, and concept synonyms in the static knowledge of letters of credit and documents. The operations for maintaining the static knowledge of letters of credit and documents should be classified according to the risk level and authorized by personnel with different permission roles.

[0108] In some other embodiments, when a third operation on the letter of credit bill review knowledge base in the bill review knowledge graph is detected, the third operation is executed to modify or add content to the letter of credit bill review knowledge base. In practical applications, business operators access the letter of credit bill review knowledge base through a browser to operate the graph management tool, so as to add or modify the content (single content or batch content) of 17 types of entity dictionaries in the letter of credit bill review knowledge base.

[0109] The content of each of the above embodiments is a related description of a method for reviewing letters of credit. Using the bill review knowledge graph, the bill entity attribute structure data is extracted from the bill to be reviewed, and the clause intention and slot data are extracted from the letter of credit to be reviewed. The bill entity attribute structure data, clause intention, and slot data are input into the bill review execution engine for review to obtain the review result, eliminating the need for manual review of letters of credit and bills, and improving the review efficiency.

[0110] In practical applications, for example Figure 7 As shown in the business architecture schematic diagram of the letter of credit bill review method provided, the letter of credit bill review method provided by the embodiments of the present invention can be applied in an international settlement business system (one of the external systems); among them, the content stored in "bill review knowledge graph storage" is the bill review knowledge graph, "attribute / relationship" and "attribute value range" are the static knowledge of letters of credit and documents, "knowledge model" is the dynamic knowledge of bill review rules, and "business knowledge base" is the letter of credit bill review knowledge base. The content stored in "bill review knowledge graph service" is all the business knowledge and bill review knowledge involved in letter of credit bill review, and the content stored in "bill review knowledge graph service" is used as the basis for the bill review execution engine to process data.

[0111] In the process of actually applying the letter of credit bill review method, the letter of credit bill review method can be divided into multiple levels, such as Figure 8 As shown in the technical architecture level diagram of the letter of credit bill review method provided; the letter of credit bill review method is divided into a basic technology layer, a knowledge data layer, a calculation layer, and an application layer; the functions of different layers are detailed in Figure 8 the content shown, which will not be elaborated here.

[0112] Corresponding to the method for examining documents of a letter of credit provided in the above embodiments of the present invention, refer to Figure 9 , an embodiment of the present invention further provides a structural block diagram of a system for examining documents of a letter of credit. The system for examining documents of a letter of credit includes: an acquisition unit 901, an extraction unit 902, and an auditing unit 903;

[0113] The acquisition unit 901 is configured to acquire the documents to be audited and the letter of credit to be audited.

[0114] The extraction unit 902 is configured to extract document entity attribute structure data from the documents to be audited and clause intent and slot data from the letter of credit to be audited based on a pre-constructed knowledge graph for document examination. The knowledge graph for document examination at least includes: static knowledge of letters of credit and documents, document examination rules, and a knowledge base for examining documents of letters of credit.

[0115] In some embodiments, the extraction unit 902 for extracting document entity attribute structure data from the documents to be audited is specifically configured to: based on the static knowledge of letters of credit and documents and the knowledge base for examining documents of letters of credit in the pre-constructed knowledge graph for document examination, extract document entity attribute structure data from the documents to be audited through an entity protocol converter.

[0116] In some embodiments, the extraction unit 902 for extracting clause intent and slot data from the letter of credit to be audited is specifically configured to: based on the static knowledge of letters of credit and documents and the knowledge base for examining documents of letters of credit in the pre-constructed knowledge graph for document examination, use a semantic analysis engine to extract clause intent and slot data from the letter of credit to be audited.

[0117] The auditing unit 903 is configured to input the document entity attribute structure data, clause intent, and slot data into a document examination execution engine, so that the document examination execution engine audits the document entity attribute structure data, clause intent, and slot data based on the document examination rules to obtain the audit results of the documents to be audited and the letter of credit to be audited.

[0118] In the embodiments of the present invention, by using the knowledge graph for document examination, document entity attribute structure data is extracted from the documents to be audited, and clause intent and slot data are extracted from the letter of credit to be audited. The document entity attribute structure data, clause intent, and slot data are input into the document examination execution engine for auditing to obtain the audit results, without the need for manual auditing of letters of credit and documents, improving the audit efficiency.

[0119] Preferably, in combination with Figure 9 the content shown, the system for examining documents of a letter of credit further includes:

[0120] A first execution unit, configured to execute a first operation to add, delete, or modify document examination rules when detecting a first operation on the document examination rules of the knowledge graph for document examination.

[0121] Preferably, in combination with Figure 9 the content shown, the letter of credit document examination system further includes:

[0122] A second execution unit, configured to execute a second operation when detecting a second operation on the static knowledge of letters of credit and documents in the document examination knowledge graph, so as to define, query, modify or delete the content in the static knowledge of letters of credit and documents.

[0123] Preferably, in combination with Figure 9 the content shown, the letter of credit document examination system further includes:

[0124] A third execution unit, configured to execute a third operation when detecting a third operation on the letter of credit document examination knowledge base in the document examination knowledge graph, so as to modify or add content in the letter of credit document examination knowledge base.

[0125] In summary, the embodiments of the present invention provide a method and system for examining letters of credit. By using a document examination knowledge graph, entity attribute structure data of documents is extracted from the documents to be examined, and clause intentions and slot data are extracted from the letters of credit to be examined. The entity attribute structure data of documents, clause intentions and slot data are input into a document examination execution engine for examination to obtain an examination result, eliminating the need for manual examination of letters of credit and documents and improving the examination efficiency.

[0126] Each embodiment in this specification is described in a progressive manner. The same or similar parts among the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for a system or system embodiment, since it is basically similar to a method embodiment, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiment. The systems and system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0127] Those skilled in the art may further realize that the units and algorithm steps of each example described in connection with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of each example have been generally described in terms of function in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the present invention.

[0128] The foregoing description of the disclosed embodiments enables those skilled in the art to practice or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for examining documents under a letter of credit, characterized in that, The method includes: Obtaining the document to be audited and the letter of credit to be audited; Based on the pre-constructed knowledge graph for document examination, extracting the document entity attribute structure data from the document to be audited, and extracting the clause intent and slot data from the letter of credit to be audited. The knowledge graph for document examination at least includes: static knowledge of letters of credit and documents, document examination rules, and a knowledge base for letter of credit document examination; the knowledge graph for document examination is used to provide the relationships between documents, between documents and letters of credit, and between documents and other documents; Inputting the document entity attribute structure data, the clause intent, and the slot data into a document examination execution engine, so that the document examination execution engine examines the document entity attribute structure data, the clause intent, and the slot data based on the document examination rules to obtain the examination results of the document to be audited and the letter of credit to be audited; The process of extracting the document entity attribute structure data from the document to be audited includes: Based on the static knowledge of letters of credit and documents and the knowledge base for letter of credit document examination in the pre-constructed knowledge graph for document examination, extracting the document entity attribute structure data from the document to be audited through an entity protocol converter; The process of extracting the clause intent and slot data from the letter of credit to be audited includes: Based on the static knowledge of letters of credit and documents and the knowledge base for letter of credit document examination in the pre-constructed knowledge graph for document examination, using a semantic analysis engine to extract the clause intent and slot data from the letter of credit to be audited.

2. The method according to claim 1, characterized in that, The method further includes: When detecting a first operation on the document examination rules in the knowledge graph for document examination, performing the first operation to add, delete, or modify the document examination rules.

3. The method according to claim 1, characterized in that, The method further includes: When detecting a second operation on the static knowledge of letters of credit and documents in the knowledge graph for document examination, performing the second operation to define, query, modify, or delete the content in the static knowledge of letters of credit and documents.

4. The method according to claim 1, characterized in that, The method further includes: When detecting a third operation on the knowledge base for letter of credit document examination in the knowledge graph for document examination, performing the third operation to modify or add the content in the knowledge base for letter of credit document examination.

5. A system for examining documents under a letter of credit, characterized in that, The system includes: An acquisition unit for obtaining the document to be audited and the letter of credit to be audited; An extraction unit for extracting the document entity attribute structure data from the document to be audited and the clause intent and slot data from the letter of credit to be audited based on the pre-constructed knowledge graph for document examination. The knowledge graph for document examination at least includes: static knowledge of letters of credit and documents, document examination rules, and a knowledge base for letter of credit document examination; the knowledge graph for document examination is used to provide the relationships between documents, between documents and letters of credit, and between documents and other documents; An examination unit for inputting the document entity attribute structure data, the clause intent, and the slot data into a document examination execution engine, so that the document examination execution engine examines the document entity attribute structure data, the clause intent, and the slot data based on the document examination rules to obtain the examination results of the document to be audited and the letter of credit to be audited; The extraction unit for extracting the document entity attribute structure data from the to-be-audited document is specifically configured to: based on the letter of credit and document static knowledge and the letter of credit document review knowledge base in the pre-constructed document review knowledge graph, extract the document entity attribute structure data from the to-be-audited document through an entity protocol converter; The extraction unit for extracting the clause intention and slot data from the to-be-audited letter of credit is specifically configured to: based on the letter of credit and document static knowledge and the letter of credit document review knowledge base in the pre-constructed document review knowledge graph, use a semantic analysis engine to extract the clause intention and slot data from the to-be-audited letter of credit.

6. The system according to claim 5, characterized in that, The system further includes: A first execution unit, configured to execute the first operation to add, delete, or modify the document review rule when detecting a first operation on the document review rule for the document review knowledge graph.

Images