Accounting document information intelligent processing system

Abstract

The invention relates to the technical field of information intelligent processing, and discloses an accounting document information intelligent processing system, which comprises the steps of collecting bill characteristics, bill two-dimensional codes, business persons, business names and business amounts of enterprise bills based on laser scanning; and judging whether the bill characteristics are qualified or not according to whether the positions, colors and fonts of the special invoice seals, the tax control codes and the two-dimensional codes in the bill characteristics are consistent with the tax department filing template or not. According to the invention, the physical characteristics of the bill are automatically compared with the filing template of the tax department through the laser scanning technology, and the invoice state of the database of the tax bureau is checked through real-time networking; according to the double verification mechanism of physical characteristics and dynamic tax data, the problems that traditional manual verification is low in efficiency and prone to missing detection are solved, the risks of false bills and repeated reimbursement are completely eradicated from the source, and a high-credibility data basis is provided for subsequent processes.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of information intelligent processing, in particular to an information intelligent processing system for accounting vouchers. BACKGROUND

[0002] The information intelligent processing system for accounting vouchers refers to a digital system that uses advanced technologies such as artificial intelligence (AI), big data, machine learning, optical character recognition (OCR), and natural language processing (NLP) to automate, intelligentize, and efficiently transform the collection, recognition, auditing, accounting, archiving, and analysis of traditional accounting vouchers (such as invoices, receipts, reimbursement forms, and accounting vouchers).

[0003] An existing patent discloses a small and micro enterprise credit-enhancing accounting data evidence storage method and system based on blockchain technology (publication number CN120257317A). This existing patent has the following core defects and deficiencies: its technical solution only focuses on the static storage and authenticity verification of accounting data (storing hash values through blockchain and storing original files through IPFS), but completely lacks dynamic verification capabilities for the business logic of the bill. Specifically: 1) It does not involve automatic comparison of bill physical characteristics (such as seals and tax control codes) with tax templates, and cannot identify counterfeit bills; 2) It does not construct a professional business database and correlation model (such as spatial trajectory and personnel qualifications), and cannot determine compliance risks of cross-regional business or professional qualifications; 3) It lacks a dynamic reimbursement baseline calculation mechanism based on position and length of service, and cannot identify abnormal amounts (such as threshold warnings of 20% above the baseline), still relying on manual auditing efficiency bottlenecks. SUMMARY

[0004] The present application provides an information intelligent processing system for accounting vouchers to solve the existing technical problems, solving the problems of low efficiency and easy to miss detection of traditional manual verification.

[0005] To solve the above technical problems, according to one aspect of the present application, more specifically, an information intelligent processing system for accounting vouchers, comprising: Collecting the bill features, bill QR codes, business people, business names, and business amounts of enterprise bills based on laser scanning; Determining whether the bill features are qualified according to whether the positions, colors, and fonts of the invoice special seal, tax control code, and two-dimensional code in the bill features are consistent with the tax department's recorded templates; Matching the business name of the bill whose bill features are qualified, which is to determine whether the business name of the bill falls into the professional business database; The information of the ticket business person falling into the professional business database is acquired, and the relevance verification of the professional business is performed based on the business person information; Based on the enterprise database, the business person's reimbursement baseline analysis is performed on the public business tickets not falling into the professional business database and the professional business tickets passing the relevance verification; According to whether the business amount of the ticket significantly exceeds the reimbursement baseline of the business person, it is judged whether the ticket audit is qualified.

[0006] Further, the ticket two-dimensional code is scanned by laser scanning to scan the invoice two-dimensional code, and the tax bureau database is connected in real time to verify whether the invoice exists, whether it is repeated reimbursement, and whether it has been red-stripped.

[0007] Further, the professional business database stores: the business name of the professional business, the abnormal degree of each ticket manual audit, the qualification grade of each ticket business person, the project experience value of each ticket business person, and the ticket presentation location, business person trajectory, and business registration location.

[0008] Further, the specific steps of establishing the relevance verification of the professional business include: 1) Determine the spatial relevance of the ticket based on the ticket presentation location, business person trajectory, and business registration location in the professional business database; 2) Determine the professional degree coefficient of the business person according to the qualification grade of the business person and the project experience value of the business person; 3) Build a spatial correlation model according to the relationship between the abnormal degree of manual audit and spatial relevance; 4) Build a professional correlation model according to the relationship between the abnormal degree of manual audit and the professional degree coefficient of the business person; 5) Based on the correlation of the spatial correlation model and the professional correlation model, a relevance verification model is established, and whether the result output by the relevance verification model exceeds the set threshold value is used to judge whether the ticket business relevance is qualified.

[0009] Further, the abnormal degree of manual audit is determined according to the time of manual audit of the ticket and the subjective evaluation of the audit staff, the ranking and proportion in the total sample data.

[0010] Further, the relevance verification model is established by fusing the two-dimensional analysis mechanism of spatial trajectory consistency and personnel professional qualification grade; The relevance verification model first calculates the spatial relevance based on the ticket presentation location, the business person historical trajectory and the geographic offset degree of the business registration location, secondly evaluates the professional ability coefficient by comprehensively considering the business person certificate validity, project experience and role weight, and finally outputs the matching quantitative value of the business person and the professional business of the ticket through a dynamic weighting method, and automatically determines whether the relevance is qualified according to a preset threshold.

[0011] Further, the specific steps of the business person's reimbursement habit baseline analysis are: 1) Based on the enterprise database, the position, length of service and professional degree of the business person in the public business ticket or professional business ticket are acquired; 2) A baseline mathematical model is constructed according to the position, length of service and professional degree of the business person, and the reimbursement baseline of the business person in the ticket is calculated by the baseline mathematical model; 3) Whether the business amount of the ticket significantly exceeds the reimbursement baseline of the business person is judged.

[0012] Further, the baseline mathematical model calculates the reimbursement baseline of the business person; The baseline mathematical model takes the enterprise unified reimbursement baseline value as the basis, superimposes the gain effect of the position weight adjusted by the adjustment factor, the nonlinear growth characteristics brought by the length of service growth, and the additional adjustment of the professional qualification coefficient, and finally outputs the personalized reimbursement amount threshold through the product relationship, providing dynamic quantitative basis for subsequent amount anomaly judgment.

[0013] Further, whether the ticket audit is qualified is determined according to whether the business amount recorded in the ticket exceeds 20% of the reimbursement baseline of the business person.

[0014] The information intelligent processing system for accounting vouchers provided by the application has the following effects compared with the prior art: 1) The application automatically compares the physical characteristics of the ticket with the tax department record template through laser scanning technology, and simultaneously verifies the invoice state of the tax bureau database in real time; the double verification mechanism of "physical characteristics + dynamic tax data" solves the problems of low efficiency and easy to miss detection in traditional manual verification, eliminates the risk of false tickets and repeated reimbursement from the source, and provides a high reliability data basis for subsequent processes.

[0015] 2) The application innovatively constructs a professional business database containing spatial information, personnel qualifications and historical abnormal data, and designs a spatial relevance model and a professional degree coefficient model. Through multi-dimensional cross analysis by the relevance verification formula, the limitation of single keyword matching is broken through, and the ticket compliance in complex scenes such as cross-regional projects is accurately identified.

[0016] 3. The present invention constructs a reimbursement baseline formula based on the enterprise database, dynamically integrates the business person's position weight, logarithmic decay of length of service and professional coefficient, realizes personalized reimbursement amount intelligent risk control, and avoids "one-size-fits-all" review.

[0017] 4. This invention integrates laser scanning acquisition, intelligent business database matching, correlation model calculation, and baseline analysis modules to form a closed-loop processing process. It automatically calculates the abnormal distance between merchant locations and business trajectories based on spatial correlation, and automatically quantifies professional coefficients and project experience, significantly reducing the manual review process. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a flow chart of the present invention; Figure 2 is a relationship diagram between the degree of abnormality and the spatial correlation S of the bill in the present invention; Figure 3 is a relationship diagram between the abnormality level and the professional level coefficient C in the present invention; Figure 4 This is a relationship diagram between the degree of abnormality and the spatial correlation S of the bill and the professional degree coefficient C in the present invention; Figure 5 Graph showing the relationship between the spatial correlation S and the distance ratio D1 / D2 in the present invention. DETAILED DESCRIPTION

[0019] In order to make the technical solution of the present invention clearer, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] Bill characteristics refer to the physical properties of the invoice stamp, tax control code and QR code; spatial correlation refers to the quantitative indicator of the geographical consistency between the merchant registration place, the bill use place and the business person's historical trajectory; the professional degree coefficient is dynamically generated by the certificate level and project role weight; the reimbursement baseline is a personalized risk control threshold that integrates position, length of service and professional qualifications. Example 1

[0021] like Figure 1 According to one aspect of the present invention, there is provided an intelligent information processing system for accounting documents, comprising: The bill features, bill QR code, business person, business name and business amount of the enterprise bill are collected based on laser scanning; the qualification of the bill features is judged based on whether the position, color and font of the invoice special stamp, tax control code and QR code in the bill features are consistent with the template filed by the tax department; the bill QR code is scanned by laser scanning to scan the invoice QR code, and is connected to the database of the State Administration of Taxation in real time to verify whether the invoice exists, whether it has been reimbursed repeatedly, and whether it has been red-offset.

[0022] The embodiment realizes the automatic comparison of the features of the bill (invoice special seal, tax control code, etc.) and the record template of the tax department by laser scanning technology, and combines real-time network verification of the State Administration of Taxation database to first deeply integrate the physical feature verification of laser scanning and dynamic tax data verification. The design not only solves the low efficiency and easy to miss detection problem of traditional manual verification of invoice authenticity, but also significantly improves the accuracy of invoice authenticity judgment through the double verification mechanism (physical features + tax data), and at the same time eliminates the risk of repeated reimbursement, red invoice, etc., and provides a high-credibility data basis for subsequent processes. Embodiment 2

[0023] As shown in Figure 1 , the bill with qualified bill features is matched with the business name, and the matching is to determine whether the business name of the bill falls into the professional business database; the professional business database stores: the business name of the professional business, the abnormal degree of each bill manual review, the qualification level of each bill business person, the project experience value of each bill business person, and the bill presentation location, business person trajectory, and merchant registered location.

[0024] The embodiment innovatively constructs a multi-dimensional professional business database, which not only contains the business name, but also integrates spatial information (bill location, merchant registered location), personnel qualification (certificate level, project experience), and historical abnormal data (manual review abnormal degree). By introducing a spatial correlation model (such as distance analysis of merchant location and business person trajectory) and dynamic professional coefficient calculation (such as weighted evaluation of qualification and experience), intelligent judgment of professional business attribution is realized. This multi-dimensional cross-verification mechanism breaks through the limitations of traditional single-keyword matching, significantly reduces the misjudgment rate, and is especially suitable for bill compliance screening in complex professional scenarios (such as cross-regional projects).

[0025] The construction of the professional business database not only contains the standard business name library, but also reversely marks high-risk business types through dynamically updated abnormal degree data (such as manual review time ranking, negative evaluation proportion). When the bill business name matches the database, the system automatically triggers the spatial trajectory and qualification verification; the unmatched bill is considered as a public business and directly enters the reimbursement baseline analysis. This double-layer shunting mechanism ensures the accuracy of professional business review, while avoiding excessive verification of public businesses.

[0026] When the bill business name matches the professional database, the spatial trajectory and qualification verification are automatically triggered; the unmatched bill is classified as a public business and directly enters the reimbursement baseline analysis module (see Figure 1 branch logic). Embodiment 3

[0027] As shown in Figures 1-5As shown, the information of the ticket service person falling into the professional service database is acquired, and the relevance verification of the professional service is performed based on the service person information; the specific steps of establishing the relevance verification of the professional service are as follows: 1) Based on the ticket presentation location, the service person trajectory and the merchant registration place in the professional service database, the spatial relevance of the ticket is determined. The calculation formula of the spatial relevance is as follows: ; In the formula, S represents the spatial relevance of the ticket presentation location, the service person trajectory and the merchant registration place. When S = 1, it means that the three locations coincide, which is an ideal case; D1 represents the distance between the ticket presentation location and the service person trajectory (if the service person trajectory passes through the location, D1 = 0); D2 represents the distance between the merchant registration place and the ticket presentation location.

[0028] Therefore, when D1 / D2 approaches 1, it means that there is a branch between the ticket presentation location and the merchant registration place; and the larger D1 / D2 is, the more obvious the abnormality of the ticket presentation is. And λ is represented as: ; In the formula, R represents the city radius of the ticket presentation location. For example: The city radius R is automatically converted from the administrative division area of the prefecture-level city, for example, R = 50 km in Suzhou; λ is a city size compensation factor, which is taken as ln2 / 60 for a super large city (such as Beijing) and ln2 / 30 for a small and medium-sized city.

[0029] As shown in Figure 5 , when the trajectory of a service person is in Shanghai, but the catering invoice is opened from Suzhou (D1 = 100 km, D2 = 10 km. The distance between Suzhou and Shanghai is 100 km, and the city radius of Suzhou is 50 km). Therefore, the spatial relevance of the ticket presentation location, the service person trajectory and the merchant registration place is: ; According to the above calculation, it can be known that the spatial relevance of the ticket presentation location, the service person trajectory and the merchant registration place is S = 0.39. Therefore, it can be known that the spatial relevance of the ticket presentation location, the service person trajectory and the merchant registration place is obviously abnormal.

[0030] The spatial relevance calculation is based on the geographical position relationship of the merchant registration place, the actual use place of the ticket and the service person trajectory. When the service person trajectory does not pass through the ticket use place, the system automatically calculates the straight-line distance (D1) between the trajectory point and the ticket location, and compares the distance with the distance (D2) between the merchant registration place and the ticket location. The geographical deviation (such as cross-city ticket) is compressed by an exponential function, so that the calculation result S value falls within the interval of 0-1, and S approaches 0, indicating that the spatial abnormal risk is extremely high.

[0031] 2) Determine the professional degree coefficient of the business person according to the qualification grade points of the business person and the project experience value of the business person. The calculation formula of the professional degree coefficient is: ; In the formula, C represents the professional degree coefficient of the business person, and the value range is 0-1; Q represents the qualification grade points of the business person (the qualification grade points are evaluated according to the certificate level of the business person and the weight of the certificate); represents the qualification full score threshold (the fixed value is 10 points); E represents the project experience value (the project experience value is evaluated by the project scale and the role weight of the business person); represents the experience full score threshold (the fixed value is 20 points); respectively represent the qualification weight and the experience weight, wherein And there is: The certificate level of the business person is: ; And the weight of the certificate is: ; For example, a business person holds a valid CFA certificate and an expired PMP certificate, so the qualification grade points of the business person are: .

[0032] Therefore, the qualification grade points of the business person are 3.0.

[0033] The project scale is: ; And the role weight is: ; For example, a business person participates in a national project and serves as a core member, and participates in a company project and serves as a principal, so there is: .

[0034] Therefore, the project experience value of the business person is 5.6.

[0035] And the qualification weight and the experience weight are respectively taken as W1=0.6 and W2=0.4, so there is: ; According to the calculation of the above formula, it can be known that the professional degree coefficient of the business person is 0.292.

[0036] The professional degree coefficient (C) is not statically assigned, but is dynamically adjusted according to the certificate timeliness (weight 0.8 within the validity period / weight 0.3 expired) and the project participation depth (weight 1.0 for the main responsible person / weight 0.3 for the participating members). The system automatically captures the certificate status and project archives in the enterprise human resource database, quantifies the personnel professional level through weighted accumulation, and avoids coefficient distortion caused by subjective evaluation.

[0037] 3) A spatial correlation model is constructed according to the relationship between the abnormal degree of manual review and the spatial correlation.

[0038] The abnormal degree of manual review is weighted by two indexes: the audit duration ranking proportion (weight 0.7) + negative evaluation frequency (weight 0.3), for example, ranking in the last 20% and negative evaluation ≥ 3 times, the abnormal degree > 80%.

[0039] For example, 100 sample data are collected, if after manual review and evaluation, the audit speed and positive evaluation of a certain bill exceed those of other 50 samples, then the abnormal degree of manual review is 50%. And the abnormal degree of manual review can be used to represent the correlation between the business person and the professional business recorded in the bill. (It can be equivalently represented because through data statistics, personnel with stronger professional ability have higher average social credibility, so their abnormal degree is smaller. Therefore, the smaller the correlation between the business person and the professional business recorded in the bill, the smaller the abnormal degree of manual review).

[0040] 4) A spatial correlation model is established according to the relationship between the abnormal degree of manual review and the spatial correlation S of the bill (as shown in Figure 2 , the red dots in the figure are the distribution of the collected 100 samples), then: (Formula 1); In the above formula 1, represents the abnormal degree of manual review of sample data after controlling the variable of professional degree coefficient C; S0 represents the baseline value of the intervention compensation of spatial correlation.

[0041] 5) A professional correlation model is constructed according to the relationship between the abnormal degree of manual review and the professional degree coefficient of the business person. A professional correlation model is established according to the relationship between the abnormal degree of manual review and the professional degree coefficient C (as shown in Figure 3 , the blue dots in the figure are the distribution of the collected 100 samples), then: (Formula 2); In the above formula 2, represents the abnormal degree of manual review of sample data after controlling the variable of spatial correlation S; b is used to control the The initial value and the abnormality degree value of manual review are close. When b=0 The initial value is approximately equal to the abnormality level value of manual review; Indicates the deviation between qualification weight and experience weight.

[0042] Among them, k is used to control and The rate of change tends to be approximately constant and is given by Figure 2 、 Figure 3 The data in can be known, when k=1 and The rates of change are equal.

[0043] 6) Based on the correlation between the spatial correlation model and the professional correlation model, a correlation verification model is established, and whether the result output by the correlation verification model exceeds the set threshold is used to determine whether the bill business correlation is qualified.

[0044] A mathematical model is established to determine the relationship between the degree of abnormality in manual review and the spatial correlation S and professional coefficient C of the bill (e.g. Figure 4 As shown, the model shows that the spatial correlation S and the professional degree coefficient C are positively correlated). Combined with the characteristic relationship of the above formula 1 and formula 2, the correlation verification model calculates the correlation between the business person and the professional business recorded in the bill as follows: ; Where h represents the correlation between the business person and the professional business recorded in the bill; S represents the spatial correlation of the bill, ranging from 0 to 1; S0 represents the baseline value for compensation for spatial correlation intervention; C represents the professional degree coefficient, ranging from 0 to 1; Indicates the deviation between qualification weight and experience weight.

[0045] Examples of this model formula are: When the enterprise's benchmark value for spatial correlation intervention compensation is S0=2. The deviation between qualification weight and experience weight is =0.8. Then the correlation verification model calculates the correlation between the business person and the professional business recorded in the bill as follows: ; In the above formula, the value of S0 is obtained by controlling the variable k in (Formula 1), and then changing the value of S0 until the output result of h in (Formula 1) is close to the 1:1 relationship with the abnormality degree value of manual review. At this time, the value of S0 is output, that is, S0=2.

[0046] In the above formula, The value of (Formula 2) is obtained by controlling the k variable (consistent with the k value in Formula 1 above), and then by changing The value of h in (Formula 2) is adjusted until the value of the abnormality level of the manual review is close to 1:1. At this time, the output is The numerical value of =0.8.

[0047] The spatial correlation of the company's bills is calculated at a certain time. When the spatial correlation of a business person's bill presentation location, business person's trajectory, and merchant registration location is S = 0.39, and the business person's professionalism coefficient is C = 0.292, then: ; According to the above calculations, the correlation between the business person of this enterprise and the professional business recorded in the bill is 58.8%. And comparing multiple sets of data: Spatial correlation S Professional degree coefficient C Correlation degree h Abnormal degree Implementation data 1 0.39 0.292 58.8% Significant abnormality Implementation data 2 0.40 0.30 59.9% Significant abnormality Implementation data 3 0.41 0.31 61.2% Further verification Implementation data 4 0.42 0.32 62.4% Further verification Implementation data 5 0.43 0.33 63.7% Further verification Implementation data 6 0.44 0.34 65.0% Further verification Implementation data 7 0.45 0.35 66.2% Further verification Implementation data 8 0.46 0.36 67.5% Further verification Implementation data 9 0.47 0.37 68.7% Further verification Implementation data 10 0.48 0.38 69.9% Further verification Implementation data 11 0.49 0.39 71.2% No obvious Table 1 Some business data and abnormality levels The data in Table 1 above show that when the sample data approaches infinity, the correlation degree h becomes the dividing line to determine whether the bill business correlation is qualified. When h < 59.9%, the bill is unqualified; when h ≥ 71.2%, the bill is qualified.

[0048] The setting of the threshold h≥71.2% is based on the cluster analysis of thousands of samples: when h≥71.2%, the manual review pass rate reaches 98.7%; when h<59.9%, the false bill detection rate is 91.4%.

[0049] In the calculation formula of correlation h, the spatial compensation reference value S0 and sensitivity parameter The determination relies on historical data regression analysis: By repeatedly adjusting parameters, the model output value h maintains a linear correlation with the degree of anomaly determined by manual review. This data-driven parameter calibration mechanism ensures that the model results are consistent with actual business experience and avoids the algorithmic black box problem. Example 4

[0050] like Figure 1 As shown, based on the enterprise database, a baseline analysis of business reimbursement habits is performed on public business bills that do not fall into the professional business database and professional business bills that have passed the relevance verification. The specific steps of the baseline analysis of business reimbursement habits are as follows: 1) Obtain the position, length of service, and professional level of the business personnel in public business bills or professional business bills based on the enterprise database; 2) According to the position, length of service, and professional degree of the business person, a benchmark mathematical model is constructed, and the reimbursement baseline of the business person in the bill is calculated by the benchmark mathematical model; 3) Whether the business amount of the bill significantly exceeds the reimbursement baseline of the business person is judged. The benchmark mathematical model calculates the reimbursement baseline of the business person, and the specific formula is: ; In the formula, B represents the reimbursement baseline; B0 represents the unified benchmark value of the whole company; P represents the position weight, and the value range is 0-1; N represents the actual working years, and the working years less than 1 year are calculated as decimals; C represents the professional degree coefficient, and C=0 when the business in the bill is a public business; and a, b, and g represent the position adjustment factor, the length of service attenuation coefficient, and the professional degree gain factor, respectively.

[0051] The value range of the position adjustment factor a is 0.2-0.5 (ordinary employee a=0.2, high-level manager a=0.5); the length of service attenuation coefficient b is fixed at 0.1; and the professional degree gain factor g is effective only when C>0, and g=0.15±0.05.

[0052] The reimbursement baseline formula uses a logarithmic function to process the length of service variable (ln(N+1)), which reflects the marginal effect of experience accumulation - the reimbursement amount of new employees grows rapidly at the beginning, and the increase of senior employees gradually flattens. The position weight (P) amplifies the management level difference through the adjustment factor a, and the professional coefficient (C) is effective only for professional businesses. This design avoids the job level or length of service discrimination problem caused by linear formula.

[0053] According to whether the business amount of the bill significantly exceeds the reimbursement baseline of the business person, it is judged whether the bill audit is qualified. Whether the bill audit is qualified is determined according to whether the business amount recorded in the bill exceeds 20% of the reimbursement baseline of the business person.

[0054] Examples of the above formula are as follows: New employee (public business). The unified benchmark value of the whole company is 800; the position adjustment factor, the length of service attenuation coefficient, and the professional degree gain factor are a=0.3, b=0.1, and g=0.15, respectively; the professional degree coefficient is 0; and the position weight is 0.2, so that: ; Middle manager (professional business). The unified benchmark value of the whole company is 800; the position adjustment factor, the length of service attenuation coefficient, and the professional degree gain factor are a=0.3, b=0.1, and g=0.15, respectively; the professional degree coefficient is 0.292; and the position weight is 0.6, so that: ; And supplement other data, so that: Position Length of service (N) Professional coefficient (C) reference value (B0) Reimbursement baseline (B) Qualified threshold (B x 1.2) Actual reimbursement amount Review results New staff 0.5 years 0 800 yuan 881 yuan 1057 yuan 900 yuan Qualified Middle manager 5 years 0.292 800 yuan 1236 yuan 1483 yuan 1300 yuan Qualified Technical director 10 years 0.5 800 yuan 1652 yuan 1982 yuan 2100 yuan Unqualified Table 2. Part of business people and based on reimbursement baseline audit data The system sets two levels of rigidity threshold to realize automatic determination: when the correlation degree h is greater than or equal to 71.2%, the professional business correlation verification is automatically passed; when the reimbursement amount exceeds the baseline value by 20%, the audit is triggered to be rejected. The threshold is derived from mass sample statistics (such as the implementation data 11 in Table 1), which ensures the transparency of the rules. The bills that do not reach the threshold are transferred to the manual review process, forming a "machine preliminary review + manual final judgment" risk control closed loop.

[0055] The above-described embodiments only express several embodiments of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent of the present application. It should be noted that for ordinary skilled persons in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of protection of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. An intelligent information processing system for accounting documents, characterized in that: Includes: Laser scanning is used to collect the bill features, bill QR code, business person, business name and business amount of corporate bills; The qualification of the bill features is determined by whether the position, color, and font of the invoice special stamp, tax control code, and QR code in the bill features are consistent with the template filed with the tax department; Matching the business name of the bill with qualified bill features, wherein the matching is to determine whether the business name of the bill falls into the professional business database; Obtain the bill business person information that falls into the professional business database, and verify the relevance of the professional business based on the business person information; Based on the enterprise database, a baseline analysis of business reimbursements is performed on public business invoices that are not included in the professional business database and professional business invoices that have passed the relevance verification. Public business invoices refer to general categories such as travel and office supplies, while professional business invoices refer to technical services and equipment procurement categories that require specific qualifications. Public business invoices and professional business invoices are distinguished by business codes in the enterprise database. Whether the bill is qualified for review is determined by whether the business amount of the bill significantly exceeds the business person's reimbursement baseline.

2. The intelligent information processing system for accounting documents according to claim 1 is characterized by: The bill QR code is scanned by laser scanning to scan the invoice QR code, and is connected to the State Administration of Taxation database in real time to verify whether the invoice exists, whether it has been reimbursed repeatedly, or whether it has been red-offset.

3. The intelligent information processing system for accounting documents according to claim 1 is characterized in that: The professional business database stores: the business name of the professional business, the abnormality level of the manual review of the bill, the qualification level of the bill business person, the project experience value of the bill business person, as well as the bill presentation location, the business person's historical trajectory, and the merchant registration place.

4. The intelligent information processing system for accounting documents according to claim 1 is characterized by: The specific steps for establishing the relevance verification of the professional business are: 1) Determine the spatial relevance of bills based on the bill presentation location, business history, and merchant registration location in the professional business database; 2) Determine the professionalism coefficient of the salesperson based on his / her qualification level and project experience; 3) Construct a spatial correlation model based on the relationship between the degree of abnormality and spatial correlation of manual review; 4) Construct a professional association model based on the relationship between the degree of abnormality of manual review and the professional degree coefficient of the business person; 5) Based on the correlation between the spatial correlation model and the professional correlation model, a correlation verification model is established, and whether the result output by the correlation verification model exceeds the set threshold is used to determine whether the bill business correlation is qualified.

5. The intelligent information processing system for accounting documents according to claim 4 is characterized in that: The degree of abnormality of the manual review is determined based on the time of manual review of the bills and the subjective evaluation of the reviewing employees, as well as the ranking and proportion in the total sample data.

6. The intelligent information processing system for accounting documents according to claim 4 is characterized in that: The correlation verification model is established by integrating the two-dimensional analysis mechanism of spatial trajectory consistency and personnel professional qualification level; The correlation verification model first calculates the spatial correlation based on the location where the bill is presented, the business person's historical trajectory and the degree of geographical offset of the merchant's registered place. Secondly, it evaluates the professional ability coefficient based on the validity of the business person's certificate, project experience and role weight. Finally, it outputs the quantitative matching value between the business person and the professional business described in the bill through a dynamic weighting method, and automatically determines whether the correlation is qualified based on the preset threshold.

7. The intelligent information processing system for accounting documents according to claim 1 is characterized by: The specific steps of the baseline analysis of business people's expense reimbursement habits are as follows: 1) Obtain the position, length of service, and professional level of the business personnel in public business bills or professional business bills based on the enterprise database; 2) Construct a benchmark mathematical model based on the salesperson's position, length of service, and professional level, and use this benchmark mathematical model to calculate the salesperson's reimbursement baseline in the bill; 3) Determine whether the business amount of the bill significantly exceeds the business person’s reimbursement baseline.

8. The intelligent information processing system for accounting documents according to claim 7 is characterized in that: The benchmark mathematical model calculates the business person's reimbursement baseline; This benchmark mathematical model is based on the company's unified reimbursement benchmark value, superimposed with the gain effect of the position weight amplified by the adjustment factor, the nonlinear growth characteristics brought about by the increase in seniority, and the additional adjustment of the professional qualification coefficient, and finally outputs the personalized reimbursement amount threshold through the product relationship, providing a dynamic quantitative basis for the subsequent judgment of abnormal amounts.

9. The intelligent information processing system for accounting documents according to claim 1 is characterized by: Whether the bill is qualified is determined by whether the business amount recorded in the bill exceeds 20% of the business person's reimbursement baseline.

Images