Finance and tax consultation service method and system based on block chain

Abstract

The invention provides a finance and tax consultation service method and system based on a block chain, and the method comprises the steps: obtaining finance and tax data of a target enterprise, splitting the finance and tax data into a structured finance voucher and an unstructured bill image, carrying out the cross verification, generating a verification result, and writing the verification result into a distributed account book; when newly added finance and tax data is detected, performing field-level semantic analysis on the structured financial voucher, extracting key business fields, and generating a tax compliance suggestion set in combination with industries and operation regions; based on the physical anti-counterfeiting feature and the verification result of the unstructured bill image, constructing a finance and tax credibility evaluation index, when the index is lower than a preset threshold value, initiating re-checking on an abnormal bill, generating an abnormal record, and performing association marking on the abnormal record and a tax compliance suggestion set; and according to the association marking result and the tax compliance suggestion set, the key business fields and the finance and tax credibility evaluation indexes are subjected to collaborative analysis, a tax optimization scheme is determined, a consultation service report is generated, and the intelligent level and compliance of enterprise finance and tax management are improved.

Description

Technical Field

[0001] The embodiments of the present application relate to the technical field of financial and tax systems, and in particular, to a financial and tax consulting service method and system based on blockchain. Background Art

[0002] In the field of enterprise financial and tax management, with the acceleration of digital transformation, the scale and complexity of enterprise financial and tax data have increased significantly. Enterprises need to process a large amount of structured financial vouchers (such as electronic invoices, bank statements) and unstructured bill images (such as scanned paper invoices), and ensure the authenticity, compliance, and traceability of these data. At the same time, enterprises also need to dynamically adapt to tax preferential policies according to their own industry characteristics and business regions to optimize tax costs and avoid compliance risks.

[0003] At present, there are already financial and tax data processing solutions based on artificial intelligence and blockchain technologies. Key information in bill images is extracted through optical character recognition (OCR) technology, and financial and tax data is verified and stored in combination with the distributed ledger characteristics of blockchain. Although the above solutions have realized the automated processing and verification of financial and tax data to a certain extent, they still have significant defects. First, the existing solutions lack in-depth analysis of the physical anti-counterfeiting features in unstructured bill images and only rely on the text information extracted by OCR for verification, making it difficult to effectively identify highly imitated forged bills. Second, when generating tax compliance suggestions, the above solutions do not fully consider the credibility differences of financial and tax data, resulting in the possibility that the proposed solutions may be generated based on low-credibility data, with relatively high compliance risks. Finally, the above solutions do not achieve the collaborative analysis of structured financial vouchers and unstructured bill images, and cannot dynamically adjust the weights of tax optimization solutions, making it difficult to meet the complex and changeable financial and tax management needs of enterprises. Summary of the Invention

[0004] The embodiments of the present application provide a financial and tax consulting service method and system based on blockchain to solve the problems of difficult multi-source heterogeneous collaborative processing of financial and tax data, insufficient ability to identify highly imitated forged bills, lack of dynamic adaptation ability of tax optimization solutions, contradiction between cross-institutional collaboration and data privacy protection, and insufficient level of financial and tax management intelligence in the prior art.

[0005] In a first aspect, the embodiments of the present application provide a financial and tax consulting service method based on blockchain, including:

[0006] Obtain the financial and tax data of a target enterprise, split the financial and tax data into structured financial vouchers and unstructured bill images, and perform cross-verification on the structured financial vouchers and unstructured bill images, generate a verification result and write it into the distributed ledger;

[0007] When it is detected that there is fiscal and tax data other than the existing fiscal and tax data in the distributed ledger, perform field-level semantic parsing on the structured financial vouchers to extract key business fields, and generate a set of tax compliance suggestions in combination with the relevance of the key business fields to the industry and business region of the target enterprise;

[0008] Based on the physical anti-counterfeiting features in the unstructured bill image and the verification result, construct a fiscal and tax credibility evaluation index for the target enterprise. When the fiscal and tax credibility evaluation index is lower than the preset threshold, initiate a joint review of the abnormal bills in the unstructured bill image to generate an abnormal record, associate and mark the abnormal record with the set of tax compliance suggestions, and generate an associated marking result;

[0009] According to the associated marking result and the set of tax compliance suggestions, conduct a collaborative analysis of the key business fields and the fiscal and tax credibility evaluation index to determine a tax optimization plan, input the tax optimization plan into the enterprise user terminal, and generate a consulting service report.

[0010] Optionally, based on the physical anti-counterfeiting features in the unstructured bill image and the verification result, construct a fiscal and tax credibility evaluation index for the target enterprise. When the fiscal and tax credibility evaluation index is lower than the preset threshold, initiate a joint review of the abnormal bills in the unstructured bill image to generate an abnormal record, associate and mark the abnormal record with the set of tax compliance suggestions, and generate an associated marking result, including:

[0011] Based on the physical anti-counterfeiting features in the unstructured bill image, use texture density distribution analysis and geometric pattern matching algorithms to extract the anti-counterfeiting logo feature vectors in the edge area of the bill, and perform weighted calculation on the anti-counterfeiting logo feature vectors with the time stamp and the number of verification nodes in the preset blockchain verification result to generate an initial credibility score for a single bill;

[0012] According to the number and invoice issuer information of the unstructured bill image, retrieve the verification records of similar historical bills in the distributed ledger. When it is detected that the bill numbers are repeated and there are invoice issuer blacklist records, dynamically correct the initial credibility score using a preset weight coefficient, generate an optimal credibility score and input it into the fiscal and tax credibility evaluation index;

[0013] When the optimal credibility score in the fiscal and tax credibility evaluation index is lower than the preset threshold, trigger a multi-role voting mechanism of tax agency nodes, audit nodes, and associated enterprise nodes in the blockchain network, and conduct a distributed consensus comparison of the signature track and transaction amount in the unstructured bill image with the corresponding key business fields in the structured financial vouchers. When more than the preset number of nodes are exceeded, it is determined as abnormal, and an abnormal record including abnormal type identification and credibility difference data is generated;

[0014] Map and match the anomaly type identifier in the anomaly record with the policy terms in the tax compliance advice set, locate the business type and amount range of the abnormal bill, associate and mark high-risk terms in the tax compliance advice set according to the business type and amount range, and generate an associated marking result based on the credibility difference data.

[0015] Optionally, when the optimal credibility score in the financial and tax credibility evaluation index is lower than the preset threshold, trigger the multi-role voting mechanism of the tax agency node, audit node and associated enterprise node in the blockchain network, and conduct a distributed consensus comparison on the signature track and transaction amount in the unstructured bill image with the corresponding key business fields in the structured financial voucher. When it exceeds the preset number of nodes, it is determined as abnormal, and an anomaly record including the anomaly type identifier and credibility difference data is generated, including:

[0016] Based on the signature track in the unstructured bill image, use the handwriting pressure feature extraction algorithm and trajectory continuity analysis algorithm to generate a signature feature vector, compare the signature feature vector with the historical signature samples in the verification result, and generate a signature anomaly identifier when the similarity is lower than the preset threshold;

[0017] Based on the transaction amount in the unstructured bill image, use the optical character recognition algorithm to extract the amount digital area, compare the amount digital area with the corresponding key business fields in the structured financial voucher, and generate an amount anomaly identifier when the numerical difference exceeds the preset tolerance range;

[0018] Use the tax agency node to verify the consistency between the bill tax rate and the policy according to the signature anomaly identifier and the amount anomaly identifier, and generate a tax compliance anomaly identifier. Use the audit node to compare the contract amount and the bank receipt amount in the structured financial voucher according to the amount anomaly identifier, and generate an audit anomaly identifier. Use the associated enterprise node to confirm the matching degree between the purchase and sales party information in the unstructured bill image and the actual business transaction record stored on the chain according to the signature anomaly identifier, and generate a business anomaly identifier;

[0019] When more than the preset number of nodes among the tax compliance anomaly identifier, audit anomaly identifier and business anomaly identifier are determined as abnormal, generate an anomaly record including the signature anomaly identifier, amount anomaly identifier and credibility difference data, where the credibility difference data is the difference between the optimal credibility score and the preset threshold.

[0020] Optionally, based on the associated marking results and the set of tax compliance suggestions, perform collaborative analysis on the critical business fields and the fiscal and tax credibility evaluation indicators to determine a tax optimization plan, input the tax optimization plan into the enterprise user terminal, and generate a consulting service report, including:

[0021] Based on the high-risk clause markings in the associated marking results, perform confidence downweighting on the policy clauses in the set of tax compliance suggestions, and combine the industry type, business operation region, and business scale data in the critical business fields to generate an initial plan set based on the rule priority sorting logic;

[0022] Based on the credibility difference data in the fiscal and tax credibility evaluation indicators, perform dynamic weight adjustment on the policy clauses in the initial plan set. When the credibility difference data exceeds the first preset difference threshold, mark the tax optimization suggestions marked with high-risk clauses as invalid. When the credibility difference data is lower than the second preset difference threshold, perform credibility weighted amplification on the deductible expense fields in the critical business fields to generate a candidate set of revised tax optimization plans;

[0023] Input the candidate set of revised tax optimization plans into a pre-set industry policy adaptation model, and based on the tax preference catalog and historical declaration data of the industry to which the target enterprise belongs, screen out tax optimization plans that meet the requirement that the fiscal and tax credibility evaluation indicators are higher than the industry benchmark value;

[0024] Input the tax optimization plan into the enterprise user terminal, and generate a consulting service report including the bill image hash value, verification node signature, and policy clause reference path.

[0025] Optionally, based on the credibility difference data in the fiscal and tax credibility evaluation indicators, perform dynamic weight adjustment on the policy clauses in the initial plan set. When the credibility difference data exceeds the first preset difference threshold, mark the tax optimization suggestions marked with high-risk clauses as invalid. When the credibility difference data is lower than the second preset difference threshold, perform credibility weighted amplification on the deductible expense fields in the critical business fields to generate a candidate set of revised tax optimization plans, including:

[0026] Based on the association relationship between the credibility difference data and the policy clauses in the initial plan set, construct a dynamic weight adjustment matrix, where the row vectors of the dynamic weight adjustment matrix are the policy clauses in the initial plan set, the column vectors are the deductible expense fields in the critical business fields, and the matrix elements are the weight coefficients of each policy clause and each deductible expense field;

[0027] When the credibility difference data exceeds a first preset difference threshold, the policy clause weight coefficient corresponding to the high-risk clause mark is set to zero, and the business type that relies on abnormal bills in the initial solution set is marked as invalid;

[0028] When the credibility difference data is lower than a second preset difference threshold, weight amplification is performed on the deductible expense field in the key business field according to the product of the absolute value of the credibility difference data and a preset weight amplification coefficient, to generate an updated weight coefficient matrix;

[0029] Based on the updated weight coefficient matrix, the policy clauses in the initial solution set are reordered, and the policy clauses with a weight coefficient of zero are eliminated to generate a revised tax optimization solution candidate set.

[0030] Optionally, obtaining the financial and taxation data of the target enterprise, splitting the financial and taxation data into structured financial vouchers and unstructured bill images, performing cross-verification on the structured financial vouchers and unstructured bill images, generating verification results and writing them into a distributed ledger, including:

[0031] Acquire financial and tax data from the target enterprise's financial system, and parse the electronic financial vouchers in the financial and tax data into structured financial vouchers, wherein the structured financial vouchers include: transaction time, transaction amount, transaction party information, and business type fields;

[0032] Performing image preprocessing on the scanned image of the paper bill in the financial and tax data to generate an unstructured bill image, wherein the unstructured bill image includes: bill number, bill issuer information and amount area;

[0033] Based on the transaction time, transaction amount and transaction party information in the structured financial voucher, a field-level comparison is performed with the bill number, bill issuer information and amount area in the unstructured bill image, and when the corresponding fields of the structured financial voucher and the unstructured bill image are consistent, a preliminary verification result is generated;

[0034] The preliminary verification result is input into a plurality of verification nodes in a preset blockchain network, and a multi-node consensus mechanism is used to perform a distributed review on the preliminary verification result. When more than a preset number of verification nodes confirm that the preliminary verification result is valid, a verification result is generated and input into a distributed ledger, wherein the plurality of verification nodes include: a tax agency node, an audit node, and an affiliated enterprise node.

[0035] Optionally, when it is detected that there is fiscal and tax data other than the existing fiscal and tax data in the distributed ledger, perform field-level semantic parsing on the structured financial vouchers to extract key business fields, and generate a set of tax compliance suggestions in combination with the relevance of the key business fields to the industry and business region of the target enterprise, including:

[0036] Monitor the distributed ledger in real time through the event listening mechanism in the blockchain network. When it is detected that there is fiscal and tax data other than the existing fiscal and tax data, trigger the field-level semantic parsing process, and use natural language processing technology to perform semantic parsing on the text fields in the fiscal and tax data other than the existing fiscal and tax data to extract key business fields. Among them, the key business fields include: industry type, business region, business scale, transaction type, and deductible expense category;

[0037] Match the industry type and business region in the key business fields with the tax preferential policies in the preset industry-region policy mapping table to generate an initial set of policy clauses;

[0038] According to the business scale and transaction type in the key business fields, screen the initial set of policy clauses, and eliminate the clauses that are not applicable to the business scale of the target enterprise to generate an intermediate set of policy clauses;

[0039] Based on the deductible expense category in the key business fields, perform priority sorting on the intermediate set of policy clauses to generate a set of tax compliance suggestions. Each policy clause in the set of tax compliance suggestions includes applicable conditions, preferential margins, and priority scores.

[0040] In a second aspect, an embodiment of the present application provides a blockchain-based fiscal and tax consulting service system, including:

[0041] A verification module, configured to obtain the fiscal and tax data of the target enterprise, split the fiscal and tax data into structured financial vouchers and unstructured bill images, and perform cross-verification on the structured financial vouchers and unstructured bill images, generate a verification result and write it into the distributed ledger;

[0042] An analysis module, configured to perform field-level semantic parsing on the structured financial vouchers to extract key business fields when it is detected that there is fiscal and tax data other than the existing fiscal and tax data in the distributed ledger, and generate a set of tax compliance suggestions in combination with the relevance of the key business fields to the industry and business region of the target enterprise;

[0043] A construction module, configured to construct a financial and tax credibility evaluation index for a target enterprise based on the physical anti-counterfeiting features in the unstructured bill image and the verification result. When the financial and tax credibility evaluation index is lower than a preset threshold, initiate a joint review of the abnormal bills in the unstructured bill image to generate an abnormal record, associate and mark the abnormal record with the tax compliance advice set, and generate an associated marking result.

[0044] An analysis module, configured to perform collaborative analysis on the key business fields and the financial and tax credibility evaluation index according to the associated marking result and the tax compliance advice set, determine a tax optimization plan, input the tax optimization plan into an enterprise user terminal, and generate a consulting service report.

[0045] In a third aspect, an embodiment of the present application provides a computing device, including a processor and a memory. A computer program is stored in the memory, and the processor is configured to run the computer program to execute any one of the methods for blockchain-based financial and tax consulting services in the first aspect.

[0046] In a fourth aspect, an embodiment of the present application provides a computer storage medium, on which computer program instructions are stored. When the computer program instructions are executed by a processor, any one of the methods for blockchain-based financial and tax consulting services in the first aspect is implemented.

[0047] In an embodiment of the present application, financial and tax data of a target enterprise is obtained, the financial and tax data is split into structured financial vouchers and unstructured bill images, and cross-verification is performed on the structured financial vouchers and the unstructured bill images to generate a verification result and write it into a distributed ledger; when it is detected that there is financial and tax data outside the existing financial and tax data in the distributed ledger, perform field-level semantic parsing on the structured financial vouchers to extract key business fields, and generate a tax compliance advice set in combination with the relevance of the key business fields to the industry and business region of the target enterprise; based on the physical anti-counterfeiting features in the unstructured bill image and the verification result, construct a financial and tax credibility evaluation index for the target enterprise. When the financial and tax credibility evaluation index is lower than a preset threshold, initiate a joint review of the abnormal bills in the unstructured bill image to generate an abnormal record, associate and mark the abnormal record with the tax compliance advice set, and generate an associated marking result; according to the associated marking result and the tax compliance advice set, perform collaborative analysis on the key business fields and the financial and tax credibility evaluation index, determine a tax optimization plan, input the tax optimization plan into an enterprise user terminal, and generate a consulting service report.

[0048] The technical solution of this application significantly improves the efficiency and accuracy of data processing by splitting the obtained fiscal and tax data into structured financial vouchers and unstructured bill images and performing cross-verification; uses the distributed ledger technology of blockchain to ensure data authenticity and traceability; generates a dynamically adapted set of tax compliance suggestions through field-level semantic parsing to help enterprises accurately match tax preferential policies; constructs a fiscal and tax credibility evaluation index based on physical anti-counterfeiting features and blockchain verification results to improve the accuracy of abnormal bill identification; finally, generates a precise and reliable tax optimization plan through collaborative analysis and generates a transparent consulting service report, comprehensively improving the intelligent level and compliance of enterprise fiscal and tax management.

[0049] Furthermore, extract bill anti-counterfeiting features through texture density distribution analysis and geometric pattern matching algorithms, generate an initial credibility score in combination with blockchain verification results, and optimize the fiscal and tax credibility evaluation index using a dynamic weight correction mechanism; perform distributed consensus comparison on abnormal bills through a multi-role voting mechanism to ensure the coordination and fairness of abnormal identification; associate and mark abnormal records with the set of tax compliance suggestions to accurately locate high-risk clauses; generate an associated marking result based on credibility difference data to provide data support for subsequent plan generation, while enhancing the transparency and auditability of fiscal and tax management, significantly improving the accuracy and reliability of abnormal bill identification and tax optimization.

[0050] These aspects or other aspects of this application will be more clearly understood in the following description of the embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of this application 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 some embodiments of this application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0052] Figure 1 It is a flowchart of a blockchain-based fiscal and tax consulting service method provided by an embodiment of this application;

[0053] Figure 2 It is a schematic structural diagram of a blockchain-based fiscal and tax consulting service system provided by an embodiment of this application;

[0054] Figure 3 It is a schematic structural diagram of a computing device provided by an embodiment of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0055] To enable those skilled in the art to better understand the solution of this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application.

[0056] In some processes described in the specification and claims of this application and the above-mentioned drawings, a plurality of operations appear in a specific order. However, it should be clearly understood that these operations may not be executed in the order in which they appear herein or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions such as "first" and "second" in this article are used to distinguish different messages, devices, modules, etc., do not represent a sequence, and do not limit that "first" and "second" are of different types.

[0057] The technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of this application.

[0058] Figure 1 A flowchart of a blockchain-based financial and tax consulting service method is provided for the embodiments of this application, as Figure 1 shown. The method includes:

[0059] Step 101: Obtain the financial and tax data of the target enterprise, split the financial and tax data into structured financial vouchers and unstructured bill images, perform cross-verification on the structured financial vouchers and unstructured bill images, generate a verification result, and write it into the distributed ledger;

[0060] In this step, structured financial vouchers refer to financial data with a clear field format, such as electronic invoices, bank statements, etc., which contain key fields such as invoice numbers, transaction amounts, and invoicing party names; unstructured bill images refer to financial data stored in image form, such as scanned paper invoices, contracts, etc., and key information needs to be extracted through image processing technology; cross-verification refers to performing a consistency comparison on the structured financial vouchers and unstructured bill images through multiple verification nodes in the blockchain network to ensure the authenticity and integrity of the data;

[0061] In this embodiment, first, the financial and tax data of the target enterprise is obtained through the data acquisition interface, and it is split into structured financial vouchers and unstructured bill images by using the data format recognition algorithm. Then, a cross-verification rule library is constructed based on the bill number consistency rule, the transaction amount matching rule, and the invoicing party information association rule. Then, multiple verification nodes in the blockchain network perform distributed verification on the financial data table and the bill image set, and generate verification results (such as the bill number consistency status, the transaction amount matching status). Finally, the verification results are bound to the hash values of the financial data table and the bill images, and a verification record with a timestamp is generated and written into the distributed ledger;

[0062] For example, a manufacturing enterprise uploaded 1,000 electronic invoices and 500 scanned paper invoices. The system split the electronic invoices into structured financial vouchers and the scanned invoices into unstructured bill images. Through multiple verification nodes in the blockchain network, the system checked the bill numbers, transaction amounts, and invoicing party information, generated verification results, and wrote them into the distributed ledger.

[0063] Step 102, when it is detected that there is financial and tax data other than the existing financial and tax data in the distributed ledger, perform field-level semantic parsing on the structured financial vouchers to extract key business fields, and generate a set of tax compliance suggestions in combination with the relevance of the key business fields to the industry and business location of the target enterprise;

[0064] In this step, field-level semantic parsing refers to performing natural language processing on the text fields in the structured financial vouchers to extract key business fields (such as industry type, business location, business scale); the set of tax compliance suggestions refers to a set of tax preferential policies generated based on the mapping table of key business fields and industry-region policies, including applicable conditions, preferential margins, and priority scores;

[0065] In this embodiment, first, the blockchain smart contract is used to monitor the data update events of the distributed ledger in real time, trigger the field-level semantic parsing process, and extract key business fields (such as industry type, business location, business scale). Then, based on the pre-set industry-region policy mapping table, the key business fields are matched with the tax preferential policies to generate an initial set of policy terms. Then, the initial set of policy terms is screened according to the business scale and transaction type, inapplicable terms are eliminated, and priority sorting is performed based on the deductible expense categories to generate a set of tax compliance suggestions;

[0066] For example, when the system detects that 100 new e-invoices have been added to the distributed ledger, it triggers the field-level semantic parsing process, extracting key business fields such as the industry type being "manufacturing", the business location being "free trade zone", and the business scale being "annual revenue below 100 million yuan". Combining with the industry-region policy mapping table, the system generates a set of tax compliance suggestions including clauses such as "additional deduction for R & D expenses" and "tax credit for environmental protection equipment".

[0067] Step 103: Based on the physical anti-counterfeiting features in the unstructured invoice image and the verification result, construct a financial and tax credibility evaluation index for the target enterprise. When the financial and tax credibility evaluation index is lower than the preset threshold, initiate a joint review of the abnormal invoices in the unstructured invoice image to generate an abnormal record, associate and mark the abnormal record with the set of tax compliance suggestions, and generate an association marking result;

[0068] In this step, the physical anti-counterfeiting features refer to anti-counterfeiting identification features such as texture density, geometric patterns, and signature tracks in the invoice image, which are used to identify forged invoices; the financial and tax credibility evaluation index refers to a scoring index generated based on physical anti-counterfeiting features and blockchain verification results, which is used to evaluate the credibility of financial and tax data; the abnormal record refers to a record containing abnormal type identifiers (such as signature abnormality, amount abnormality) and credibility difference data, which is used to mark high-risk invoices

[0069] First, adopt texture density distribution analysis and geometric pattern matching algorithms to extract the physical anti-counterfeiting features in the invoice image and generate an initial credibility score. Then, according to invoice number duplicate detection and blacklist records of the invoicing party, dynamically correct the initial credibility score using a preset weight coefficient to generate an optimal credibility score. Then, when the optimal credibility score is lower than the preset threshold, trigger a multi-role voting mechanism to conduct a joint review of the abnormal invoices and generate an abnormal record (such as containing a "signature abnormality" identifier and credibility difference data);

[0070] For example, when the system detects that the initial credibility score of a certain invoice is 60 points (lower than the preset threshold of 70 points), it triggers a multi-role voting mechanism. The tax agency node, audit node, and associated enterprise node respectively verify the signature track, transaction amount, and invoicing party information, and finally determine that the invoice is abnormal, generating an abnormal record containing a "signature abnormality" identifier.

[0071] Step 104: According to the association marking result and the set of tax compliance suggestions, conduct a collaborative analysis of the key business fields and the financial and tax credibility evaluation index to determine a tax optimization plan, input the tax optimization plan into the enterprise user terminal, and generate a consulting service report;

[0072] In this step, collaborative analysis is used to combine the associated marking results and the set of tax compliance suggestions, comprehensively analyze the key business fields and the financial and tax credibility evaluation indicators, and generate accurate tax optimization solutions; the consulting service report refers to a comprehensive report containing tax optimization solutions, invoice image hash values, verification node signatures, and policy clause reference paths, which is used to support the enterprise's financial and tax management decisions;

[0073] In this embodiment, first, according to the high-risk clause markings in the associated marking results, the set of tax compliance suggestions is de-weighted to generate an initial solution set. Then, based on the credibility difference data, the dynamic weight of the initial solution set is adjusted to eliminate high-risk clauses and generate a candidate set of revised tax optimization solutions. Then, the candidate set is input into the industry policy adaptation model to screen out the final tax optimization solutions that meet the industry benchmark values. Finally, the final solutions are pushed to the enterprise user terminal through an encrypted channel, and a consulting service report is generated;

[0074] For example, the system de-weights the "business travel expense reimbursement" clause according to the "signature anomaly" identification in the abnormal record, and combines the credibility difference data to generate a candidate set of revised tax optimization solutions. Through the industry policy adaptation model, the system screens out "additional deduction of R & D expenses" and "tax credit for environmental protection equipment" as the final solutions and generates a consulting service report to be pushed to the enterprise user terminal.

[0075] In summary, through the above steps, the embodiment of the present application realizes the efficient processing and collaborative verification of financial and tax data, significantly improves the data authenticity and traceability; generates a dynamically adapted set of tax compliance suggestions through field-level semantic parsing and industry and regional policy matching; constructs financial and tax credibility evaluation indicators based on physical anti-counterfeiting features and blockchain verification results, accurately identifies abnormal invoices and generates associated marking results; finally, generates accurate and reliable tax optimization solutions through collaborative analysis and generates a transparent consulting service report, comprehensively improving the efficiency, reliability, and intelligent level of the enterprise's financial and tax management.

[0076] In enterprise financial and tax management, the authenticity verification of unstructured bill images (such as scanned paper invoices) is a key problem. Traditional methods mainly rely on manual review or simple OCR technology, which are difficult to effectively identify highly imitated forged bills and lack a quantitative evaluation mechanism for bill credibility. In addition, existing technologies do not fully utilize the distributed characteristics of blockchain, resulting in low efficiency in identifying and reviewing abnormal bills. Based on this, to solve the above problems, in some embodiments, according to step 103, based on the physical anti-counterfeiting features in the unstructured bill image and the verification result, a financial and tax credibility evaluation index for the target enterprise is constructed. When the financial and tax credibility evaluation index is lower than the preset threshold, a joint review is initiated for the abnormal bills in the unstructured bill image to generate an abnormal record, and the abnormal record is associated and marked with the set of tax compliance suggestions, and an associated marking result is generated, including:

[0077] Step 201: Based on the physical anti-counterfeiting features in the unstructured bill image, use texture density distribution analysis and geometric pattern matching algorithms to extract the anti-counterfeiting logo feature vectors in the edge area of the bill, and perform weighted calculation on the anti-counterfeiting logo feature vectors, the timestamp and the number of verification nodes in the preset blockchain verification result to generate an initial credibility score for a single bill;

[0078] In this step, the physical anti-counterfeiting features refer to anti-counterfeiting logo features such as texture density, geometric patterns, and signature trajectories in the bill image, which are used to identify forged bills; the anti-counterfeiting logo feature vectors refer to the digital feature representations of the edge area of the bill extracted by texture density distribution analysis and geometric pattern matching algorithms, which are used to quantify the anti-counterfeiting features; the initial credibility score refers to the credibility score of a single bill generated by weighted calculation based on the anti-counterfeiting logo feature vectors, the timestamp and the number of verification nodes in the blockchain verification result;

[0079] In this embodiment, first, use the texture density distribution analysis algorithm to extract the texture features (such as local texture complexity) in the edge area of the bill image, and detect the fixed patterns (such as invoice code box, amount column position) in the bill layout through the geometric pattern matching algorithm to generate anti-counterfeiting logo feature vectors. Then, perform weighted calculation on the anti-counterfeiting logo feature vectors, the timestamp (used to verify the timeliness of the bill) and the number of verification nodes (used to evaluate the extensiveness of the verification) in the blockchain verification result to generate an initial credibility score. Finally, the initial credibility score is used as the preliminary result of the bill authenticity verification for subsequent dynamic correction and multi-role voting mechanisms.

[0080] Step 202: According to the number and the invoicing party information of the unstructured bill image, retrieve the verification records of similar historical bills in the distributed ledger. When detecting duplicate bill numbers and invoicing party blacklist records, dynamically correct the initial credibility score using a preset weight coefficient, generate an optimal credibility score, and input it into the fiscal and tax credibility evaluation indicators;

[0081] In this step, the verification records of similar historical bills refer to the verification results of historical bills stored in the distributed ledger with the same bill number and invoicing party information as the current bill; the optimal credibility score refers to the final credibility score generated by dynamically correcting the initial credibility score using a preset weight coefficient based on duplicate bill number detection and invoicing party blacklist records; the fiscal and tax credibility evaluation indicators refer to the indicators for evaluating the credibility of fiscal and tax data constructed based on the optimal credibility score, and are used for subsequent abnormal bill identification and tax optimization plan generation;

[0082] In this embodiment, first, according to the number and the invoicing party information of the unstructured bill image, retrieve the verification records of similar historical bills in the distributed ledger. If duplicate bill numbers or invoicing party blacklist records are detected, dynamically correct the initial credibility score using a preset weight coefficient (such as deducting 20% of the weight for duplicate numbers and 50% of the weight for blacklist records), and generate an optimal credibility score. Finally, the optimal credibility score is input into the fiscal and tax credibility evaluation indicators to evaluate the credibility of the bill.

[0083] Step 203: When the optimal credibility score in the fiscal and tax credibility evaluation indicators is lower than the preset threshold, trigger the multi-role voting mechanism of the tax agency node, the audit node, and the associated enterprise node in the blockchain network, and conduct a distributed consensus comparison on the signature track and the transaction amount in the unstructured bill image with the corresponding key business fields in the structured financial vouchers. When more than the preset number of nodes is reached, it is determined as abnormal, and an abnormal record including the abnormal type identifier and the credibility difference data is generated;

[0084] In this step, the multi-role voting mechanism refers to an abnormal bill review mechanism jointly participated by the tax agency node, the audit node, and the associated enterprise node in the blockchain network, ensuring the fairness of abnormal identification through distributed consensus comparison; the abnormal type identifier refers to the specific problem type of the abnormal bill, such as signature abnormality, amount abnormality, etc.; the credibility difference data refers to the difference between the optimal credibility score and the preset threshold, and is used to quantify the risk level of the abnormal bill;

[0085] In this embodiment, when the optimal credibility score in the fiscal and tax credibility evaluation index is lower than the preset threshold (such as 70 points), a multi-role voting mechanism in the blockchain network is triggered. The tax agency node verifies the authenticity of the signature track, the audit node compares the consistency of the transaction amount with the key business fields, and the associated enterprise node confirms the relevance of the invoicing party information. If more than the preset nodes (such as more than half) determine it to be abnormal, an abnormal record containing an abnormal type identifier (such as "signature anomaly") and credibility difference data (such as a difference of 10%) is generated.

[0086] Step 204: Map and match the abnormal type identifier in the abnormal record with the policy terms in the tax compliance advice set, locate the business type and amount range of the abnormal bill, associate and mark the high-risk terms in the tax compliance advice set according to the business type and amount range, and generate an associated marking result based on the credibility difference data;

[0087] In this step, the high-risk terms refer to the policy terms in the tax compliance advice set associated with the abnormal bill, which may have relatively high compliance risks; the associated marking result refers to the marking result generated based on the abnormal type identifier and credibility difference data, and is used to guide the generation of the tax optimization plan.

[0088] In this embodiment, first, map and match the abnormal type identifier in the abnormal record with the policy terms in the tax compliance advice set to locate the business type and amount range of the abnormal bill. Then, associate and mark the high-risk terms in the tax compliance advice set according to the business type and amount range, and generate an associated marking result based on the credibility difference data. Finally, the associated marking result is used for the dynamic adjustment of the subsequent tax optimization plan.

[0089] For example, a manufacturing enterprise uploads 100 electronic invoices and 50 scanned paper invoices. First, the system extracts the anti-counterfeiting logo feature vectors of the paper invoices through texture density distribution analysis and geometric pattern matching algorithms, and generates an initial credibility score in combination with the blockchain verification results. Then, the system detects that the number of one of the bills is repeated with the historical record, and the invoicing party is on the blacklist. Therefore, the system dynamically corrects the initial credibility score using the preset weight coefficient to generate an optimal credibility score (such as 60 points). Since the optimal credibility score is lower than the preset threshold (70 points), the system triggers a multi-role voting mechanism. The tax agency node, the audit node, and the associated enterprise node respectively verify the signature track, the transaction amount, and the invoicing party information, and finally determine that the bill is abnormal, generating an abnormal record containing an "abnormal signature" identifier and credibility difference data (such as a difference of 10%). Subsequently, the system maps and matches the "abnormal signature" identifier with the "business travel reimbursement clause" in the tax compliance advice set, and associates and marks this clause as high risk in the set. Finally, the system generates a tax optimization plan based on the associated marking results and pushes it to the enterprise user terminal.

[0090] In summary, through the above steps, this embodiment realizes the extraction of physical anti-counterfeiting features of unstructured bill images and the calculation of credibility scores, significantly improving the accuracy of bill authenticity verification; through the dynamic weight correction mechanism and the multi-role voting mechanism, it ensures the fairness and reliability of abnormal bill identification; through the associated marking of high-risk clauses, it helps enterprises avoid potential compliance risks; and finally generates accurate and reliable tax optimization plans, comprehensively improving the intelligent level and compliance of enterprise financial and tax management.

[0091] In financial and tax management, the identification of abnormal bills is a key issue, especially the accurate identification of highly imitated forged bills. Existing technologies usually rely on a single data source or simple rules for abnormal judgment, making it difficult to effectively identify complex forgery behaviors. In addition, existing solutions lack a cross-institutional collaboration mechanism, resulting in insufficient fairness and transparency in abnormal identification. Based on this, as another embodiment, according to step 203, when the optimal credibility score in the financial and tax credibility evaluation index is lower than the preset threshold, a multi-role voting mechanism of the tax agency node, the audit node, and the associated enterprise node in the blockchain network is triggered to conduct a distributed consensus comparison on the signature track and transaction amount in the unstructured bill image and the corresponding key business fields in the structured financial vouchers. When more than the preset number of nodes is reached, it is determined to be abnormal, and an abnormal record containing an abnormal type identifier and credibility difference data is generated, including:

[0092] Step 301: Based on the signature trajectory in the unstructured bill image, use the handwriting pressure feature extraction algorithm and the trajectory continuity analysis algorithm to generate a signature feature vector. Compare the similarity between the signature feature vector and the historical signature samples in the verification result. When the similarity is lower than the preset threshold, generate a signature anomaly flag.

[0093] In this step, the signature trajectory refers to the handwriting trajectory of the signature in the bill image, including features such as handwriting pressure and stroke continuity. The signature feature vector refers to the digital feature representation generated by the handwriting pressure feature extraction algorithm and the trajectory continuity analysis algorithm, which is used to quantify the authenticity of the signature trajectory. The signature anomaly flag refers to the flag generated when the similarity between the signature feature vector and the historical signature sample is lower than the preset threshold, which is used to mark the signature anomaly.

[0094] In this embodiment, first, use the handwriting pressure feature extraction algorithm to analyze the change of handwriting pressure in the signature trajectory (such as pressure peak value, pressure distribution uniformity), and detect the number of breakpoints and continuity index of the signature trajectory (such as the proportion of continuous stroke length) through the trajectory continuity analysis algorithm to generate a signature feature vector. Then, compare the similarity between the signature feature vector and the historical signature samples in the verification result (such as cosine similarity). If the similarity is lower than the preset threshold (such as 70%), generate a signature anomaly flag. Finally, the signature anomaly flag is used for anomaly determination in the subsequent multi-role voting mechanism.

[0095] Step 302: Based on the transaction amount in the unstructured bill image, use the optical character recognition algorithm to extract the amount digital area, and compare the numerical value of the amount digital area with the corresponding key business field in the structured financial voucher. When the numerical difference exceeds the preset tolerance range, generate an amount anomaly flag.

[0096] In this step, the transaction amount refers to the amount digital area displayed in the bill image, usually presented in text form. The amount anomaly flag is the flag generated when the numerical difference between the amount digital area and the corresponding key business field in the structured financial voucher exceeds the preset tolerance range, which is used to mark the amount anomaly.

[0097] In this embodiment, first use the optical character recognition (OCR) algorithm to extract the amount digital area (such as the total invoice amount column) in the bill image and identify the amount value. Then, compare the recognized amount with the corresponding key business field (such as the contract amount) in the structured financial voucher. If the numerical difference exceeds the preset tolerance range (such as ±5%), generate an amount anomaly flag. Finally, the amount anomaly flag is used for anomaly determination in the subsequent multi-role voting mechanism.

[0098] Step 303: The tax agency node verifies the consistency between the bill tax rate and the policy based on the signature anomaly flag and the amount anomaly flag, and generates a tax compliance anomaly flag. The audit node compares the contract amount in the structured financial voucher with the bank receipt amount based on the amount anomaly flag, and generates an audit anomaly flag. The related enterprise node confirms the matching of the purchase and sales party information in the unstructured bill image with the actual business transaction records stored on the chain based on the signature anomaly flag, and generates a business anomaly flag;

[0099] In this step, the tax compliance anomaly flag refers to the flag generated after the tax agency node verifies the consistency between the bill tax rate and the policy, and is used to mark tax compliance anomalies; the audit anomaly flag refers to the flag generated after the audit node compares the contract amount with the bank receipt amount, and is used to mark audit anomalies; the business anomaly flag refers to the flag generated after the related enterprise node confirms the matching of the purchase and sales party information with the actual business transaction records stored on the chain, and is used to mark business anomalies;

[0100] In this embodiment, first, the tax agency node verifies the consistency between the bill tax rate and the policy (such as whether the tax rate of the special VAT invoice is 13%) based on the signature anomaly flag and the amount anomaly flag, and generates a tax compliance anomaly flag. Then, the audit node compares the contract amount in the structured financial voucher with the bank receipt amount based on the amount anomaly flag. If the difference exceeds the tolerance range, an audit anomaly flag is generated. Next, the related enterprise node confirms whether the purchase and sales party information in the unstructured bill image matches the actual business transaction records stored on the chain based on the signature anomaly flag. If they do not match, a business anomaly flag is generated. Finally, the tax compliance anomaly flag, the audit anomaly flag, and the business anomaly flag are used for anomaly determination in the multi-role voting mechanism.

[0101] Step 304: When more than a preset number of nodes among the tax compliance anomaly flag, the audit anomaly flag, and the business anomaly flag are determined to be abnormal, an anomaly record including the signature anomaly flag, the amount anomaly flag, and the credibility difference data is generated, where the credibility difference data is the difference between the optimal credibility score and the preset threshold;

[0102] In this step, the anomaly record refers to the record including the signature anomaly flag, the amount anomaly flag, and the credibility difference data, and is used to mark the specific problems and risk levels of abnormal bills; the credibility difference data refers to the difference between the optimal credibility score and the preset threshold, and is used to quantify the risk level of abnormal bills;

[0103] In this embodiment, when more than a preset number (such as more than half) of the nodes among the tax compliance exception flag, the audit exception flag, and the business exception flag are determined to be abnormal, an exception record including a signature exception flag, an amount exception flag, and credibility difference data is generated. The credibility difference data is calculated by the difference between the optimal credibility score and the preset threshold (such as the difference is 10%), and is used for the dynamic adjustment of the subsequent tax optimization plan. Finally, the exception record is used for the associated marking of high-risk clauses and the generation of tax optimization plans.

[0104] For example, the system first uses a handwriting pressure feature extraction algorithm and a trajectory continuity analysis algorithm to extract the signature trajectory features and generate a signature feature vector. By comparing with historical signature samples, it is found that the similarity is only 65% (lower than the preset threshold of 70%), and a signature exception flag is generated. Then, the system uses the OCR algorithm to extract the amount digital area in the bill image, identifies the amount as 100,000 yuan, but the difference from the contract amount (95,000 yuan) in the structured financial voucher exceeds the tolerance range (±5%), and an amount exception flag is generated. Subsequently, the tax agency node verifies the consistency between the bill tax rate and the policy, and finds that the tax rate meets the requirements, and no tax compliance exception flag is generated; the audit node compares the contract amount with the bank receipt amount and finds that the difference exceeds the tolerance range, and an audit exception flag is generated; the associated enterprise node confirms that the purchase and sales party information does not match the actual business transaction record stored on the chain, and a business exception flag is generated. Since the audit exception flag and the business exception flag exceed the preset number (such as more than half), the system generates an exception record including a signature exception flag, an amount exception flag, and credibility difference data (such as the difference is 10%). Finally, the exception record is used for the associated marking of high-risk clauses and the generation of tax optimization plans.

[0105] In summary, through the above steps, this embodiment realizes the in-depth analysis of the signature trajectory and the multi-source comparison of the transaction amount, significantly improving the accuracy of abnormal bill recognition; through the multi-role voting mechanism, it ensures the fairness and transparency of abnormal recognition; through the generation of exception records and the quantitative feedback of credibility difference data, it provides accurate data support for the generation of subsequent tax optimization plans, comprehensively improving the intelligent level and compliance of enterprise financial and tax management.

[0106] Since the prior art usually relies on a static rule library or a single data source to generate tax optimization plans, and does not fully consider the credibility differences and industry and regional characteristics of financial and tax data, resulting in the plan being possibly generated based on low-credibility data and lacking dynamic adaptation ability. To solve the above problems, in some embodiments, according to step 104, based on the associated marking result and the tax compliance advice set, the key business fields and the financial and tax credibility evaluation indicators are analyzed collaboratively to determine the tax optimization plan, and the tax optimization plan is input into the enterprise user terminal, and a consulting service report is generated, including:

[0107] Step 401: Based on the high-risk clause markings in the associated marking results, perform confidence downweighting on the policy clauses in the tax compliance advice set, and combine the industry type, business location, and business scale data in the key business fields to generate an initial solution set based on the rule priority sorting logic;

[0108] In this step, the high-risk clause marking refers to the policy clauses in the tax compliance advice set associated with abnormal bills, which may have relatively high compliance risks; the confidence downweighting process is used to reduce the priority score of relevant policy clauses according to the high-risk clause markings, so as to reduce their weight in the tax optimization solution; the initial solution set refers to the candidate set of tax optimization solutions generated based on the rule priority sorting logic, including applicable conditions, preferential margins, and priority scores;

[0109] In this embodiment, first, based on the high-risk clause markings in the associated marking results, perform confidence downweighting on the relevant clauses in the tax compliance advice set, and then combine the industry type, business location, and business scale data in the key business fields to generate an initial solution set based on the rule priority sorting logic. Finally, the initial solution set is used as the input for subsequent dynamic weight adjustment to generate a revised candidate set of tax optimization solutions.

[0110] Step 402: Based on the credibility difference data in the financial and tax credibility evaluation indicators, perform dynamic weight adjustment on the policy clauses in the initial solution set. When the credibility difference data exceeds the first preset difference threshold, mark the tax optimization suggestions with high-risk clause markings as invalid. When the credibility difference data is lower than the second preset difference threshold, perform credibility weighted amplification on the deductible expense fields in the key business fields to generate a revised candidate set of tax optimization solutions;

[0111] In this step, the credibility difference data refers to the difference between the optimal credibility score and the preset threshold, which is used to quantify the risk level of abnormal bills; the dynamic weight adjustment refers to the process of adjusting the weights of the policy clauses in the initial solution set according to the credibility difference data to optimize the accuracy of the tax optimization solution; the revised candidate set of tax optimization solutions refers to the candidate set of tax optimization solutions generated after dynamic weight adjustment, including optimization suggestions supported by high-credibility data.

[0112] In this embodiment, first, based on the credibility difference data in the fiscal and tax credibility evaluation indicators (such as a difference of 10%), dynamic weight adjustment is performed on the policy clauses in the initial solution set. When the credibility difference data exceeds the first preset difference threshold (such as >40%), invalidation marks are made on the tax optimization suggestions marked for high-risk clauses; when the credibility difference data is lower than the second preset difference threshold (such as <10%), credibility weighting amplification is performed on the deductible expense fields (such as "R & D expenses") in the key business fields. Finally, a candidate set of revised tax optimization solutions is generated for subsequent screening by the industry policy adaptation model.

[0113] Step 403: Input the candidate set of the revised tax optimization solutions into a preset industry policy adaptation model, and based on the tax preference catalog of the industry to which the target enterprise belongs and the historical declaration data, screen out the tax optimization solutions that meet the fiscal and tax credibility evaluation indicators higher than the industry benchmark value;

[0114] In this step, the industry policy adaptation model refers to a model constructed based on the tax preference catalog of the industry to which the target enterprise belongs and the historical declaration data, and is used to screen tax optimization solutions that conform to the industry characteristics; the industry benchmark value refers to the average value of the historical fiscal and tax credibility evaluation indicators of the industry to which the target enterprise belongs, and is used to screen tax optimization solutions with high credibility;

[0115] In this embodiment, first, the candidate set of the revised tax optimization solutions is input into a preset industry policy adaptation model. Then, based on the tax preference catalog of the industry to which the target enterprise belongs and the historical declaration data, screen out the tax optimization solutions that meet the fiscal and tax credibility evaluation indicators higher than the industry benchmark value (such as 85 points). Finally, generate tax optimization solutions that conform to the industry characteristics and have high credibility for pushing to the enterprise user terminal.

[0116] Step 404: Input the tax optimization solutions into the enterprise user terminal, and generate a consulting service report including the bill image hash value, verification node signature, and policy clause reference path;

[0117] In this step, the consulting service report refers to a comprehensive report including tax optimization solutions, bill image hash values, verification node signatures, and policy clause reference paths, and is used to support enterprise fiscal and tax management decisions; the bill image hash value refers to the unique identifier of the bill image, and is used to ensure the immutability and traceability of the data; the verification node signature refers to the digital signature of the blockchain node participating in the fiscal and tax data verification, and is used to ensure the transparency and auditability of the verification process;

[0118] In this embodiment, first, the final tax optimization plan is pushed to the enterprise user terminal through an encrypted channel. Then, a consultation service report is generated, which includes the hash value of the bill image, the signature of the verification node, and the reference path of the policy terms. Finally, the consultation service report provides transparent and traceable financial and tax management support for the enterprise, enhancing the credibility of the enterprise's financial and tax management.

[0119] For example, the system first performs a confidence down-weighting process on the relevant terms in the tax compliance advice set according to the high-risk term marks in the associated marking results, and combines the industry type, business location, and business scale data in the key business fields to generate an initial plan set. Then, based on the credibility difference data (such as a difference of 10%) in the financial and tax credibility evaluation indicators, a dynamic weight adjustment is performed on the policy terms in the initial plan set to generate a revised candidate set of tax optimization plans. Next, the candidate set is input into the industry policy adaptation model to screen out tax optimization plans that meet the financial and tax credibility evaluation indicators higher than the industry benchmark value (such as 85 points). Finally, the system pushes the tax optimization plan to the enterprise user terminal and generates a consultation service report that includes the hash value of the bill image, the signature of the verification node, and the reference path of the policy terms.

[0120] Since the prior art usually relies on a static rule base or a single data source to generate tax optimization plans and does not fully consider the credibility differences of financial and tax data, resulting in the possibility that the plans may be generated based on low-credibility data and there are relatively high compliance risks. Based on this, as another embodiment, according to step 402, a dynamic weight adjustment is performed on the policy terms in the initial plan set based on the credibility difference data in the financial and tax credibility evaluation indicators. When the credibility difference data exceeds the first preset difference threshold, an invalidation mark is made for the tax optimization suggestions marked with high-risk terms. When the credibility difference data is lower than the second preset difference threshold, the credibility of the deductible expense field in the key business fields is weighted and amplified to generate a revised candidate set of tax optimization plans, including:

[0121] Step 501: Construct a dynamic weight adjustment matrix based on the association relationship between the credibility difference data and the policy terms in the initial plan set. Among them, the row vector of the dynamic weight adjustment matrix is the policy terms in the initial plan set, the column vector is the deductible expense field in the key business fields, and the matrix element is the weight coefficient of each policy term and each deductible expense field.

[0122] In this step, the dynamic weight adjustment matrix refers to a matrix constructed based on the association relationship between the credibility difference data and the policy terms in the initial plan set, which is used to dynamically adjust the weight coefficients of the policy terms; the row vector refers to the policy terms in the initial plan set;

[0123] The column vector refers to the deductible expense field in the key business fields; the weight coefficient refers to the associated weight of each policy clause with each deductible expense field, which is used to quantify the importance of the policy clause;

[0124] In this embodiment, first, based on the association relationship between the credibility difference data (such as the difference is 10%) and the policy clauses in the initial solution set, a dynamic weight adjustment matrix is constructed. The row vector of the matrix is the policy clause, the column vector is the deductible expense field, and the matrix element is the weight coefficient of each policy clause with each deductible expense field. Finally, the dynamic weight adjustment matrix is used for subsequent high-risk clause invalidation marking and deductible expense field weight amplification.

[0125] Step 502: When the credibility difference data exceeds the first preset difference threshold, set the weight coefficient of the policy clause corresponding to the high-risk clause mark to zero, and perform an invalidation mark on the business types that rely on abnormal bills in the initial solution set;

[0126] In this step, the high-risk clause mark refers to the policy clause associated with the abnormal bill, which may have a relatively high compliance risk; the invalidation mark means setting the weight coefficient of the policy clause corresponding to the high-risk clause mark to zero and removing it from the initial solution set;

[0127] In this embodiment, when the credibility difference data exceeds the first preset difference threshold (such as > 40%), the system sets the weight coefficient of the policy clause corresponding to the high-risk clause mark to zero, and performs an invalidation mark on the business types that rely on abnormal bills in the initial solution set (such as "business trip reimbursement"). Finally, the clauses with invalidation marks are removed to ensure that the tax optimization solution is generated based on high-credibility data.

[0128] Step 503: When the credibility difference data is lower than the second preset difference threshold, amplify the weight of the deductible expense field in the key business fields according to the product of the absolute value of the credibility difference data and the preset weight amplification coefficient, and generate an updated weight coefficient matrix;

[0129] In this step, weight amplification means increasing the weight coefficient of the deductible expense field according to the product of the absolute value of the credibility difference data and the preset weight amplification coefficient; the updated weight coefficient matrix refers to the weight coefficient matrix generated after weight amplification, which is used to reorder the policy clauses;

[0130] In this embodiment, when the credibility difference data is lower than the second preset difference threshold (such as < 10%), the system amplifies the weight of the deductible expense field in the key business fields according to the product of the absolute value of the credibility difference data and the preset weight amplification coefficient (such as 2 times), and finally generates an updated weight coefficient matrix for subsequent reordering of the policy clauses.

[0131] Step 504: Based on the updated weight coefficient matrix, reorder the policy clauses in the initial solution set, and eliminate the policy clauses with zero weight coefficients to generate a candidate set of revised tax optimization solutions;

[0132] In this embodiment, first, based on the updated weight coefficient matrix, the policy clauses in the initial solution set are sorted in descending order of weight coefficients. Then, the policy clauses with zero weight coefficients are eliminated to generate a candidate set of revised tax optimization solutions. Finally, the revised candidate set is used for the screening of the subsequent industry policy adaptation model.

[0133] For example, the system first constructs a dynamic weight adjustment matrix based on the correlation between the credibility difference data and the policy clauses in the initial solution set. Then, the system detects that the credibility difference data of a certain bill is 50%, sets the weight coefficient of the corresponding "business trip expense reimbursement clause" of the high-risk clause to zero, and marks the business types relying on abnormal bills as invalid. Next, the system detects that the credibility difference data of another bill is 5%, amplifies the weight of "R & D investment amount", and generates an updated weight coefficient matrix. Finally, the system reorders the policy clauses in the initial solution set based on the updated weight coefficient matrix, eliminates the invalid clauses, and generates a candidate set of revised tax optimization solutions.

[0134] To solve the problem that the existing technology relies on manual review or a single data source for verification, making it difficult to efficiently process large-scale, multi-source heterogeneous financial and tax data, resulting in low data processing efficiency. In some embodiments, according to step 101, the financial and tax data of the target enterprise is obtained, the financial and tax data is split into structured financial vouchers and unstructured bill images, and cross-verification is performed on the structured financial vouchers and unstructured bill images, and the verification results are written into the distributed ledger, including:

[0135] Step 601: Obtain the financial and tax data from the financial system of the target enterprise, and parse the electronic financial vouchers in the financial and tax data into structured financial vouchers, where the structured financial vouchers include: transaction time, transaction amount, transaction party information, and business type fields;

[0136] In this step, the financial and tax data refers to the electronic financial vouchers and scanned images of paper bills stored in the enterprise's financial system, including data such as transaction time, transaction amount, transaction party information, business type fields, bill numbers, invoice issuer information, and amount areas; the structured financial vouchers refer to the financial data with clear field formats generated by parsing the electronic financial vouchers, including transaction time, transaction amount, transaction party information, and business type fields, usually stored in the form of tables or databases;

[0137] In this embodiment, first, obtain fiscal and tax data from the financial system of the target enterprise, including electronic financial vouchers and scanned images of paper bills. Then, use a data parsing algorithm to parse the electronic financial vouchers into structured financial vouchers, and extract fields such as transaction time, transaction amount, transaction party information, and business type. Finally, the generated structured financial vouchers are used for subsequent field-level comparison and preliminary verification.

[0138] Step 602: Perform image preprocessing on the scanned images of paper bills in the fiscal and tax data to generate unstructured bill images, where the unstructured bill images include: bill numbers, issuer information, and amount regions;

[0139] In this step, the unstructured bill image refers to the digital representation of the scanned image of the paper bill generated through image preprocessing, including key information such as bill numbers, issuer information, and amount regions; image preprocessing refers to performing processing such as noise reduction, binarization, and edge detection on the scanned image of the paper bill to extract clear bill numbers, issuer information, and amount regions;

[0140] In this embodiment, first perform image preprocessing on the scanned images of paper bills in the fiscal and tax data, including noise reduction, binarization, and edge detection. Then, use an optical character recognition (OCR) algorithm to extract bill numbers, issuer information, and amount regions to generate unstructured bill images. Finally, the generated unstructured bill images are used for subsequent field-level comparison and preliminary verification.

[0141] Step 603: Based on the transaction time, transaction amount, and transaction party information in the structured financial vouchers, perform field-level comparison with the bill numbers, issuer information, and amount regions in the unstructured bill images. When the corresponding fields of the structured financial vouchers and the unstructured bill images are consistent, generate a preliminary verification result;

[0142] In this step, field-level comparison refers to performing consistency comparison between the transaction time, transaction amount, and transaction party information in the structured financial vouchers and the bill numbers, issuer information, and amount regions in the unstructured bill images to verify the authenticity and consistency of the data; the preliminary verification result refers to the consistency verification result generated after field-level comparison and is used for subsequent distributed review;

[0143] In this embodiment, first, based on the transaction time, transaction amount, and transaction party information in the structured financial vouchers, perform field-level comparison with the bill numbers, issuer information, and amount regions in the unstructured bill images. Then, judge whether the corresponding fields are consistent through a preset consistency rule. If they are consistent, generate a preliminary verification result. Finally, the preliminary verification result is used for subsequent distributed review in the blockchain network.

[0144] Step 604: Input the preliminary verification result into multiple verification nodes in a preset blockchain network, and use a multi-node consensus mechanism to conduct distributed review on the preliminary verification result. When more than a preset number of verification nodes confirm that the preliminary verification result is valid, generate a verification result and input it into the distributed ledger. Among them, the multiple verification nodes include: tax agency nodes, audit nodes, and associated enterprise nodes:

[0145] In this step, the multi-node consensus mechanism refers to a distributed review mechanism jointly participated by multiple verification nodes (such as tax agency nodes, audit nodes, and associated enterprise nodes) in the blockchain network, which is used to ensure the fairness and transparency of the verification result; the distributed ledger refers to a distributed database in the blockchain network that stores the verification result, ensuring the immutability and traceability of the data;

[0146] In this embodiment, first input the preliminary verification result into multiple verification nodes in a preset blockchain network, and then use the multi-node consensus mechanism to conduct distributed review on the preliminary verification result. Each node respectively verifies the consistency of the invoice number, transaction amount, and invoice issuer information. If more than a preset number of verification nodes confirm that the preliminary verification result is valid, generate a verification result and write it into the distributed ledger. Finally, the verification result is used for the construction of subsequent fiscal and tax credibility evaluation indicators and the identification of abnormal invoices.

[0147] For example, the system first obtains fiscal and tax data from the financial system, parses the electronic invoice into a structured financial voucher, and extracts fields such as transaction time, transaction amount, transaction party information, and business type. Then, perform image preprocessing on the scanned paper invoice to generate an unstructured invoice image, and extract the invoice number, invoice issuer information, and amount area. Then, the system conducts field-level comparison based on the transaction time, transaction amount, and transaction party information in the structured financial voucher with the invoice number, invoice issuer information, and amount area in the unstructured invoice image to generate a preliminary verification result. Finally, the system inputs the preliminary verification result into multiple verification nodes in the blockchain network, uses the multi-node consensus mechanism for distributed review, generates a verification result and writes it into the distributed ledger.

[0148] To solve the problem that the existing technology usually relies on a static rule library or a single data source to generate tax compliance suggestions, without fully considering the diversity of the actual business scenarios of enterprises, resulting in insufficient accuracy and practicality of the proposed solutions. In some embodiments, according to what is described in step 102, when it is detected that there is fiscal and tax data outside the existing fiscal and tax data in the distributed ledger, perform field-level semantic parsing on the structured financial voucher to extract key business fields, and combine the relevance of the key business fields with the industry and business region of the target enterprise to generate a set of tax compliance suggestions, including:

[0149] Step 701: Monitor the distributed ledger in real time through the event monitoring mechanism in the blockchain network. When it detects tax and financial data other than the existing tax and financial data, trigger the field-level semantic parsing process, and use natural language processing technology to perform semantic parsing on the text fields in the tax and financial data other than the existing tax and financial data, and extract key business fields. Among them, the key business fields include: industry type, business operation area, business scale, transaction type, and deductible expense category;

[0150] In this step, the event monitoring mechanism refers to the mechanism in the blockchain network that monitors the update of distributed ledger data in real time, and is used to detect new tax and financial data and trigger subsequent processing processes; field-level semantic parsing refers to using natural language processing technology to perform semantic analysis on the text fields in tax and financial data and extract key business fields; key business fields refer to the core information related to the enterprise operation scenario extracted from tax and financial data, and are used to generate tax compliance suggestions;

[0151] In this embodiment, first, monitor the distributed ledger in real time through the event monitoring mechanism in the blockchain network. When new tax and financial data is detected, trigger the field-level semantic parsing process. Then, use natural language processing technology to perform semantic parsing on the text fields in the new tax and financial data, extract key business fields, and finally the extracted key business fields are used for subsequent industry-region policy matching and tax compliance suggestion generation.

[0152] Step 702: Match the industry type and business operation area in the key business fields with the tax preferential policies in the preset industry-region policy mapping table to generate an initial policy clause set;

[0153] In this step, the industry-region policy mapping table: refers to the pre-set mapping relationship table between industry type and business operation area and tax preferential policies, and is used to match tax preferential policies suitable for enterprises; the initial policy clause set: refers to the set of tax preferential policies generated by matching industry type and business operation area, including applicable conditions, preferential range, and priority scores;

[0154] In this embodiment, first, match the industry type and business operation area in the key business fields with the tax preferential policies in the preset industry-region policy mapping table, then generate an initial policy clause set, and finally the initial policy clause set is used for subsequent screening and priority ranking.

[0155] Step 703: According to the business scale and transaction type in the key business fields, screen the initial policy clause set, eliminate the clauses that are not applicable to the business scale of the target enterprise, and generate an intermediate policy clause set;

[0156] In this step, the intermediate policy clause set refers to the set of tax preference policies generated by screening the initial policy clause set according to the business scale and transaction type, excluding inapplicable clauses; the business scale screening refers to excluding inapplicable clauses according to the business scale of the enterprise; the transaction type screening refers to retaining relevant preferential policies according to the transaction type of the enterprise.

[0157] In this embodiment, first, according to the business scale and transaction type in the key business fields, the initial policy clause set is screened, then inapplicable clauses are excluded, relevant preferential policies are retained, and finally an intermediate policy clause set is generated.

[0158] Step 704: Based on the deductible expense categories in the key business fields, prioritize the intermediate policy clause set to generate a tax compliance advice set, where each policy clause in the tax compliance advice set includes applicable conditions, preferential margins, and priority scores.

[0159] In this step, the tax compliance advice set refers to the set of tax preference policies generated after prioritization, and each policy clause includes applicable conditions, preferential margins, and priority scores.

[0160] In this embodiment, first, based on the deductible expense categories in the key business fields, the policy clauses in the intermediate policy clause set are prioritized, then a tax compliance advice set is generated, and each policy clause includes applicable conditions, preferential margins, and priority scores. Finally, the tax compliance advice set is used for the generation of subsequent tax optimization plans.

[0161] For example, the system detects new fiscal and tax data through the event listening mechanism in the blockchain network, triggers the field-level semantic parsing process, extracts the key business fields. Then, the system matches the industry type and business operation region with the preset industry-region policy mapping table to generate an initial policy clause set. Then the system screens the initial policy clause set according to the business scale and transaction type, excludes inapplicable clauses, and generates an intermediate policy clause set. Finally, the system prioritizes the intermediate policy clause set based on the deductible expense categories, generates a tax compliance advice set, and pushes it to the enterprise user terminal.

[0162] Figure 2 The structure diagram of a blockchain-based fiscal and tax consulting service system is provided for the embodiments of this application, as Figure 2 shown. The system includes:

[0163] A verification module 21, configured to obtain the fiscal and tax data of a target enterprise, split the fiscal and tax data into structured financial vouchers and unstructured bill images, and perform cross-verification on the structured financial vouchers and unstructured bill images, generate a verification result, and write it into the distributed ledger.

[0164] A parsing module 22, configured to, when detecting tax and financial data other than the existing tax and financial data in the distributed ledger, perform field-level semantic parsing on the structured financial vouchers to extract key business fields, and generate a set of tax compliance suggestions by combining the relevance of the key business fields to the industry and business region of the target enterprise;

[0165] A construction module 23, configured to construct an evaluation index of the tax and financial credibility of the target enterprise based on the physical anti-counterfeiting features in the unstructured bill images and the verification results. When the tax and financial credibility evaluation index is lower than a preset threshold, initiate a joint review of the abnormal bills in the unstructured bill images to generate abnormal records, associate and mark the abnormal records with the set of tax compliance suggestions, and generate an associated marking result;

[0166] An analysis module 24, configured to perform collaborative analysis on the key business fields and the tax and financial credibility evaluation index according to the associated marking result and the set of tax compliance suggestions, determine a tax optimization plan, input the tax optimization plan into the enterprise user terminal, and generate a consulting service report.

[0167] Figure 2 The described blockchain-based tax and financial consulting service system can execute Figure 1 The described blockchain-based tax and financial consulting service method in the illustrated embodiment, the implementation principle and technical effects will not be elaborated. For the blockchain-based tax and financial consulting service system in the above embodiment, the specific manners in which each module and unit perform operations have been described in detail in the embodiment related to the method, and will not be elaborated here.

[0168] In a possible design, Figure 2 The blockchain-based tax and financial consulting service system in the illustrated embodiment can be implemented as a computing device, such as Figure 3 shown, the computing device may include a storage component 31 and a processing component 32;

[0169] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32.

[0170] The processing component 32 is used to obtain the financial and tax data of the target enterprise, split the financial and tax data into structured financial vouchers and unstructured bill images, perform cross-verification on the structured financial vouchers and unstructured bill images, generate a verification result and write it into the distributed ledger; when it is detected that there is financial and tax data other than the existing financial and tax data in the distributed ledger, perform field-level semantic parsing on the structured financial vouchers to extract key business fields, combine the relevance of the key business fields with the industry and operating region of the target enterprise, and generate a set of tax compliance suggestions; based on the physical anti-counterfeiting features in the unstructured bill images and the verification result, construct a financial and tax credibility evaluation index for the target enterprise. When the financial and tax credibility evaluation index is lower than the preset threshold, initiate a joint review of the abnormal bills in the unstructured bill images to generate an abnormal record, associate and mark the abnormal record with the set of tax compliance suggestions, and generate an associated marking result; according to the associated marking result and the set of tax compliance suggestions, perform collaborative analysis on the key business fields and the financial and tax credibility evaluation index to determine a tax optimization plan, input the tax optimization plan into the enterprise user terminal, and generate a consulting service report.

[0171] Among them, the processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component may also be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components for executing the above method.

[0172] The storage component 31 is configured to store various types of data to support the operation of the terminal. The storage component can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0173] Of course, the computing device may also necessarily include other components, such as input / output interfaces, display components, communication components, etc.

[0174] The input / output interface provides an interface between the processing component and the peripheral interface module, and the above peripheral interface module may be an output device, an input device, etc.

[0175] The communication component is configured to facilitate wired or wireless communication between the computing device and other devices, etc.

[0176] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform. In this case, the computing device can refer to a cloud server. The above-mentioned processing components, storage components, etc. can be basic server resources rented or purchased from a cloud computing platform.

[0177] The embodiment of the present application also provides a computer storage medium storing a computer program, which when executed by a computer can implement the above-mentioned Figure 1 method for blockchain-based fiscal and tax consulting services shown in the embodiments.

[0178] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0179] The device 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 labor.

[0180] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0181] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A financial and tax consulting service method based on blockchain, characterized in that: include: Obtain the financial and tax data of the target enterprise, split the financial and tax data into structured financial vouchers and unstructured bill images, perform cross-verification on the structured financial vouchers and unstructured bill images, generate verification results and write them into the distributed ledger; When it is detected that financial and tax data other than existing financial and tax data appears in the distributed ledger, field-level semantic parsing is performed on the structured financial voucher to extract key business fields, and a tax compliance recommendation set is generated based on the correlation between the key business fields and the industry and business region of the target enterprise; Based on the physical anti-counterfeiting features in the unstructured bill image and the verification result, construct a financial and tax credibility evaluation index of the target enterprise; when the financial and tax credibility evaluation index is lower than a preset threshold, initiate a joint review of the abnormal bills in the unstructured bill image to generate abnormal records, associate and mark the abnormal records with the tax compliance suggestion set, and generate an association marking result; According to the association marking result and the tax compliance suggestion set, the key business fields and the financial and tax credibility evaluation indicators are collaboratively analyzed to determine a tax optimization plan, input the tax optimization plan into the enterprise user terminal, and generate a consulting service report.

2. The method according to claim 1, characterized in that Based on the physical anti-counterfeiting features in the unstructured bill image and the verification result, a financial and tax credibility evaluation index of the target enterprise is constructed. When the financial and tax credibility evaluation index is lower than a preset threshold, a joint review is initiated for the abnormal bills in the unstructured bill image to generate abnormal records, the abnormal records are associated with the tax compliance suggestion set, and an associated marking result is generated, including: Based on the physical anti-counterfeiting features in the unstructured bill image, the anti-counterfeiting mark feature vector in the edge area of ​​the bill is extracted by using texture density distribution analysis and geometric pattern matching algorithm, and the anti-counterfeiting mark feature vector is weightedly calculated with the timestamp and the number of verification nodes in the preset blockchain verification result to generate an initial credibility score for a single bill; According to the number and issuer information of the unstructured bill image, the verification records of similar historical bills are retrieved in the distributed ledger. When duplicate bill numbers and issuer blacklist records are detected, the initial credibility score is dynamically corrected using a preset weight coefficient to generate an optimal credibility score and input it into the financial and tax credibility evaluation index; When the optimal credibility score in the financial and tax credibility evaluation index is lower than the preset threshold, the multi-role voting mechanism of the tax agency node, audit node and associated enterprise node in the blockchain network is triggered, and the signature track and transaction amount in the unstructured bill image are compared with the corresponding key business fields in the structured financial voucher through distributed consensus. When it exceeds the preset node, it is judged as an abnormality, and an abnormal record containing the abnormality type identification and credibility difference data is generated; The exception type identifier in the exception record is mapped and matched with the policy clauses in the tax compliance suggestion set, the business type and amount range of the abnormal bill are located, and high-risk clauses are associated and marked in the tax compliance suggestion set according to the business type and amount range, and an associated marking result is generated based on the credibility difference data.

3. The method according to claim 2, characterized in that When the optimal credibility score in the financial and tax credibility evaluation index is lower than the preset threshold, the multi-role voting mechanism of the tax agency node, audit node and associated enterprise node in the blockchain network is triggered, and the signature track and transaction amount in the unstructured bill image are compared with the corresponding key business fields in the structured financial voucher through distributed consensus. When it exceeds the preset node, it is judged as an abnormality, and an abnormal record containing the abnormal type identification and credibility difference data is generated, including: Based on the signature track in the unstructured bill image, a handwriting pressure feature extraction algorithm and a track continuity analysis algorithm are used to generate a signature feature vector, and the signature feature vector is compared with the historical signature sample in the verification result for similarity. When the similarity is lower than a preset threshold, a signature abnormality mark is generated; Based on the transaction amount in the unstructured bill image, an optical character recognition algorithm is used to extract the amount digital area, and the amount digital area is numerically compared with the corresponding key business field in the structured financial voucher. When the numerical difference exceeds a preset tolerance range, an amount abnormality mark is generated; The tax agency node is used to verify the consistency of the bill tax rate with the policy based on the signature abnormality mark and the amount abnormality mark, and generate a tax compliance abnormality mark. The audit node is used to compare the contract amount in the structured financial voucher with the bank receipt amount based on the amount abnormality mark, and generate an audit abnormality mark. The associated enterprise node is used to confirm the matching of the buyer and seller information in the unstructured bill image with the actual business transaction records stored on the chain based on the signature abnormality mark, and generate a business abnormality mark; When more than a preset number of nodes among the tax compliance exception mark, audit exception mark and business exception mark are determined to be abnormal, an exception record is generated including a signature exception mark, an amount exception mark and credibility difference data, wherein the credibility difference data is the difference between the optimal credibility score and a preset threshold.

4. The method according to claim 1, characterized in that: According to the association tag result and the tax compliance suggestion set, the key business fields and the financial and tax credibility evaluation indicators are collaboratively analyzed to determine a tax optimization plan, input the tax optimization plan into the enterprise user terminal, and generate a consulting service report, including: According to the high-risk clause tags in the association tag results, the policy clauses in the tax compliance suggestion set are downgraded in terms of confidence, and combined with the industry type, operating region and business scale data in the key business field, an initial solution set based on the rule priority sorting logic is generated; Based on the credibility difference data in the financial and tax credibility evaluation index, dynamic weight adjustment is performed on the policy clauses in the initial solution set. When the credibility difference data exceeds a first preset difference threshold, the tax optimization suggestions marked as high-risk clauses are marked as invalid. When the credibility difference data is lower than a second preset difference threshold, the credibility of the deductible expense field in the key business field is weighted and amplified to generate a revised tax optimization solution candidate set. Input the revised tax optimization solution candidate set into a preset industry policy adaptation model, and screen out tax optimization solutions that meet the financial and tax credibility evaluation index higher than the industry benchmark value based on the tax preferential catalog of the industry to which the target enterprise belongs and historical declaration data; The tax optimization plan is input into the enterprise user terminal, and a consulting service report including the bill image hash value, verification node signature and policy clause reference path is generated.

5. The method according to claim 4, characterized in that Based on the credibility difference data in the financial and tax credibility evaluation index, dynamic weight adjustment is performed on the policy clauses in the initial solution set. When the credibility difference data exceeds a first preset difference threshold, the tax optimization suggestions marked as high-risk clauses are marked as invalid. When the credibility difference data is lower than a second preset difference threshold, the deductible expense field in the key business field is credibility-weighted amplified to generate a revised tax optimization solution candidate set, including: Based on the association between the credibility difference data and the policy clauses in the initial solution set, a dynamic weight adjustment matrix is ​​constructed, wherein the row vectors of the dynamic weight adjustment matrix are the policy clauses in the initial solution set, the column vectors are the deductible expense fields in the key business fields, and the matrix elements are the weight coefficients of each policy clause and each deductible expense field; When the credibility difference data exceeds a first preset difference threshold, the policy clause weight coefficient corresponding to the high-risk clause mark is set to zero, and the business type that relies on abnormal bills in the initial solution set is marked as invalid; When the credibility difference data is lower than a second preset difference threshold, weight amplification is performed on the deductible expense field in the key business field according to the product of the absolute value of the credibility difference data and a preset weight amplification coefficient, to generate an updated weight coefficient matrix; Based on the updated weight coefficient matrix, the policy clauses in the initial solution set are reordered, and the policy clauses with a weight coefficient of zero are eliminated to generate a revised tax optimization solution candidate set.

6. The method according to claim 1, characterized in that Obtain the financial and tax data of the target enterprise, split the financial and tax data into structured financial vouchers and unstructured bill images, perform cross-verification on the structured financial vouchers and unstructured bill images, generate verification results and write them into a distributed ledger, including: Acquire financial and tax data from the target enterprise's financial system, and parse the electronic financial vouchers in the financial and tax data into structured financial vouchers, wherein the structured financial vouchers include: transaction time, transaction amount, transaction party information, and business type fields; Performing image preprocessing on the scanned image of the paper bill in the financial and tax data to generate an unstructured bill image, wherein the unstructured bill image includes: bill number, bill issuer information and amount area; Based on the transaction time, transaction amount and transaction party information in the structured financial voucher, a field-level comparison is performed with the bill number, bill issuer information and amount area in the unstructured bill image, and when the corresponding fields of the structured financial voucher and the unstructured bill image are consistent, a preliminary verification result is generated; The preliminary verification result is input into a plurality of verification nodes in a preset blockchain network, and a multi-node consensus mechanism is used to perform a distributed review on the preliminary verification result. When more than a preset number of verification nodes confirm that the preliminary verification result is valid, a verification result is generated and input into a distributed ledger, wherein the plurality of verification nodes include: a tax agency node, an audit node, and an affiliated enterprise node.

7. The method according to claim 1, characterized in that When it is detected that financial and tax data other than existing financial and tax data appears in the distributed ledger, field-level semantic parsing is performed on the structured financial voucher to extract key business fields, and a tax compliance recommendation set is generated based on the correlation between the key business fields and the industry and business region of the target enterprise, including: The distributed ledger is monitored in real time through the event monitoring mechanism in the blockchain network. When fiscal and tax data other than the existing fiscal and tax data is detected, the field-level semantic parsing process is triggered, and the text fields in the fiscal and tax data other than the existing fiscal and tax data are semantically parsed using natural language processing technology to extract key business fields, wherein the key business fields include: industry type, operating region, business scale, transaction type, and deductible expense category; Matching the industry type and business region in the key business field with the tax preferential policies in the preset industry and region policy mapping table to generate an initial policy clause set; According to the business scale and transaction type in the key business field, the initial policy clause set is screened, clauses that are not applicable to the target enterprise's business scale are eliminated, and an intermediate policy clause set is generated; Based on the deductible expense categories in the key business fields, the intermediate policy clause set is prioritized to generate a tax compliance suggestion set, wherein each policy clause in the tax compliance suggestion set includes applicable conditions, preferential ranges, and priority scores.

8. A financial and tax consulting service system based on blockchain, characterized in that: include: A verification module, used to obtain the financial and tax data of the target enterprise, split the financial and tax data into structured financial vouchers and unstructured bill images, perform cross-verification on the structured financial vouchers and unstructured bill images, generate verification results and write them into the distributed ledger; A parsing module, configured to perform field-level semantic parsing on the structured financial voucher to extract key business fields when detecting that financial and tax data other than existing financial and tax data appear in the distributed ledger, and generate a set of tax compliance recommendations based on the correlation between the key business fields and the industry and business region of the target enterprise; A construction module, for constructing a financial and tax credibility evaluation index of a target enterprise based on the physical anti-counterfeiting features in the unstructured bill image and the verification result, and when the financial and tax credibility evaluation index is lower than a preset threshold, initiating a joint review of abnormal bills in the unstructured bill image to generate abnormal records, associating and marking the abnormal records with the tax compliance suggestion set, and generating an associated marking result; The analysis module is used to collaboratively analyze the key business fields and the financial and tax credibility evaluation indicators based on the association marking results and the tax compliance recommendation set, determine the tax optimization plan, input the tax optimization plan into the enterprise user terminal, and generate a consulting service report.

9. A computing device, characterized in that It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a blockchain-based financial and tax consulting service method as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that: A computer program is stored, and when the computer program is executed by a computer, a financial and tax consulting service method based on blockchain as described in any one of claims 1 to 7 is implemented.

Images