A smart identification system and method for preventing duplicate receipts

Abstract

This invention discloses an intelligent identification system and method for preventing duplicate receipts, belonging to the field of data recognition technology. The invention collects digital signatures and text information from electronic receipts, adds timestamps, and constructs historical reimbursement behavior; extracts writing features from the digital signatures to construct receipts to be detected; filters receipts within a preset interval according to timestamps, calculates writing feature similarity, and marks suspected duplicate receipts; sorts the suspected duplicate receipts to form a detection sequence, compares it with the historical reimbursement sequence, and marks it as passed if a matching sequence exists; otherwise, the first receipt is automatically reimbursed, and the rest are transferred to manual review; collects the results of manual review and updates the corresponding receipt reimbursement history. The system includes modules for data collection and construction, data feature extraction, similarity judgment, classification and comparison processing, and result recording and updating. This invention improves the accuracy of duplicate receipt identification, significantly reduces the amount of invalid manual review, and improves the efficiency of reimbursement review.

Description

Technical Field

[0001] This invention relates to the field of big data analytics, specifically to an intelligent identification system and method for preventing duplicate receipts. Background Technology

[0002] Driven by both digital transformation and the deepening of cross-regional business operations, invoices, as core vouchers for financial reimbursement and compliance auditing, are experiencing increasingly complex circulation scenarios. The widespread adoption of electronic invoices has solved the problem of inefficient paper invoice delivery, but it may also lead to the same invoice being submitted to multiple departments simultaneously, causing the system to repeatedly identify it as duplicate reimbursement and requiring multiple manual reviews, thus reducing the efficiency of invoice review. Summary of the Invention

[0003] The purpose of this invention is to provide an intelligent identification system and method for preventing duplicate receipts, in order to solve the problems raised in the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: an intelligent identification method for preventing duplicate receipts, the intelligent identification method comprising the following steps:

[0005] Step S1: Collect the digital signature and text information of the electronic invoice and attach a timestamp and historical reimbursement behavior, wherein the historical reimbursement behavior is the reimbursement sequence of all invoices with the same digital signature;

[0006] Step S2: Clean and standardize the collected digital signatures and text information, extract the writing features of the digital signatures, and construct the invoice to be tested;

[0007] Step S3: Extract all tickets to be detected within a preset range before and after the timestamp, collect all writing features, and mark tickets to be detected with a similarity greater than a preset threshold as suspected duplicate tickets.

[0008] Step S4: Extract all suspected duplicate identification tickets, classify them according to writing characteristics, sort the suspected duplicate identification tickets according to timestamps to form a detection sequence, query whether there is a ticket reimbursement sequence with the same detection sequence in the historical reimbursement behavior. If there is, mark it as passed. If not, reimburse the first suspected duplicate identification ticket and transfer the subsequent suspected duplicate identification tickets to manual review.

[0009] Step S5: Collect the results of manual review and update the corresponding invoice reimbursement history.

[0010] Furthermore, the specific steps of step S1 are as follows:

[0011] Step S1-1: Collect the digital signature and text information that comes with the electronic ticket, wherein the text information includes all text information on the electronic ticket;

[0012] Step S1-2: Generate a timestamp based on the real-time submission time of the electronic ticket, and bind and associate the timestamp with the digital signature and text information of the corresponding electronic ticket to form a preliminary data set;

[0013] Steps S1-3: Extract all past expense reimbursement data corresponding to a certain digital signature, organize them in the order of reimbursement time to form a complete expense reimbursement sequence, and construct historical reimbursement behavior.

[0014] Furthermore, the specific steps of step S2 are as follows:

[0015] Step S2-1: Organize the digital signature according to the system's preset format and convert the text information into the preset format.

[0016] Step S2-2: Extract writing features from the regularized digital signature, including the stroke logic and structural layout of the signature;

[0017] Step S2-3: The extracted writing features, converted text information, timestamps, and historical reimbursement behavior are associated and integrated to form the invoice to be tested.

[0018] Furthermore, the specific steps of step S3 are as follows:

[0019] Step S3-1: Extract a ticket to be tested. Based on its timestamp and the system's preset timestamp filtering range, construct a preset time interval for the ticket to be tested. From all the tickets to be tested, extract all the tickets whose timestamps fall within the preset time interval to form a temporary set of tickets to be tested.

[0020] Step S3-2: Extract the digital signature writing features of each document to be tested from the temporary set of documents to be tested, and form a temporary writing feature set from all the extracted writing features.

[0021] Step S3-3: Using a preset feature similarity calculation method, perform pairwise comparisons of all writing features in the temporary writing feature set to obtain the similarity value between each pair of writing features. Compare the calculated similarity value with the system's preset threshold. If the similarity value of two sets of writing features is greater than the system's preset threshold, then the tickets to be detected associated with these two sets of features are jointly marked as suspected duplicate tickets. All information marked as suspected duplicate tickets is summarized and organized to form a list of suspected duplicate tickets.

[0022] Furthermore, the specific steps of step S4 are as follows:

[0023] Step S4-1: Classify all suspected duplicate identification tickets according to the extracted writing features, and group suspected duplicate identification tickets with completely identical writing features into the same category;

[0024] Step S4-2: For suspected duplicate identification tickets within each category, sort them according to their bound timestamp information, and associate and bind the digital signature and standardized text information of each suspected duplicate identification ticket with the sequence to be detected to form the sequence to be detected for each category.

[0025] Step S4-3: Extract the invoice reimbursement sequence from the historical reimbursement behavior associated with the digital signature corresponding to the sequence to be detected. Compare the sequence to be detected with the extracted historical reimbursement sequence. If the comparison result shows that there is a completely identical invoice reimbursement sequence, then all suspected duplicate invoices in this category are marked as passed. If the comparison result shows that there is no completely identical invoice reimbursement sequence, then the suspected duplicate invoice ranked first will be automatically reimbursed, and the remaining invoices will be transferred to the manual review process. Record the comparison result and the corresponding invoice association information.

[0026] Furthermore, the specific steps of step S5 are as follows:

[0027] Step S5-1: Collect the comparison results generated by the manual review process, and summarize the comparison results with the digital signature, standardized text information and timestamp already bound to the invoice to form a record document;

[0028] Step S5-2: Update the expense reimbursement history of the invoice corresponding to the digital signature according to the record document.

[0029] Furthermore, the intelligent recognition system includes a data acquisition and construction module, a data feature extraction module, a similarity judgment module, a classification and comparison processing module, and a result recording and updating module;

[0030] The data acquisition and construction module is used to collect the digital signatures and text information of electronic invoices and generate timestamps, while constructing a historical reimbursement sequence for all invoices corresponding to the same digital signature. The data feature extraction module is used to clean and standardize the collected digital signatures and text information, extract the writing features of the digital signatures, and associate them with timestamps and historical reimbursement behaviors to construct invoices to be detected. The similarity judgment module is used to filter invoices within a preset time interval based on the timestamps of the invoices to be detected, and to mark suspected duplicate invoices by calculating the similarity of writing features. The classification and comparison processing module is used to classify suspected duplicate invoices according to writing features, sort them according to timestamps to form a sequence to be detected, and perform corresponding processing after comparing them with historical reimbursement sequences. The result recording and updating module is used to collect the comparison results generated by the manual review process, summarize the results with the invoice information, and update the historical reimbursement records of the corresponding digital signatures.

[0031] The output of the data acquisition and construction module is electrically connected to the input of the data feature extraction module; the output of the data feature extraction module is electrically connected to the input of the screening similarity determination module; the output of the screening similarity determination module is electrically connected to the input of the classification comparison processing module; and the output of the classification comparison processing module is electrically connected to the input of the result recording and updating module.

[0032] Furthermore, the data acquisition construction module includes an information acquisition unit and a historical behavior construction unit;

[0033] The information collection unit is used to collect the digital signature of the electronic invoice and the text information consisting of all the text on the invoice, and generate a timestamp associated with the invoice based on the real-time submission time of the invoice; the historical behavior construction unit is used to extract all past invoice reimbursement data corresponding to a certain digital signature, and organize them into a complete invoice reimbursement sequence according to the reimbursement time.

[0034] Furthermore, the data feature extraction module includes a data regularization unit and a feature extraction unit;

[0035] The data normalization unit is used to normalize the collected digital signatures according to the system's preset format, and at the same time convert the text information into the system's preset format; the feature extraction unit is used to extract the stroke logic and structural layout writing features from the normalized digital signatures, and integrate them with the converted text information, timestamps and historical reimbursement behavior to form the invoice to be tested.

[0036] The similarity determination module includes a time-based filtering unit and a similarity determination unit;

[0037] The time filtering unit is used to construct a preset time interval based on the timestamp of a certain ticket to be detected and the preset timestamp filtering range of the system, and then extract tickets whose timestamps fall within the interval from all tickets to be detected to form a temporary ticket set to be detected; the similarity determination unit is used to perform pairwise comparison calculation of the writing features of all tickets in the temporary ticket set to be detected using a preset feature similarity calculation method, and after comparing the similarity value with a preset threshold, mark suspected duplicate identification tickets and summarize them.

[0038] Furthermore, the classification and comparison processing module includes a ticket classification and sorting unit and a comparison result processing unit;

[0039] The invoice classification and sorting unit is used to classify all suspected duplicate invoices according to the extracted writing features, group invoices with the same writing features into the same category, and then sort the invoices in the category according to the timestamp and associate information to form a sequence to be detected; the comparison result processing unit is used to extract the historical reimbursement sequence corresponding to the digital signature of the sequence to be detected, compare the two, if they match, mark the invoices in the category as passed, if they do not match, the first invoice is automatically reimbursed, and the rest are transferred to manual review and the results are recorded.

[0040] The result recording and updating module includes a result collection unit and a record updating unit;

[0041] The result collection unit is used to collect the invoice comparison results generated by the manual review process, and summarize the results with the digital signature, standardized text information and timestamp bound to the invoice to form a record document; the record update unit is used to update the invoice reimbursement history of the digital signature corresponding to the invoice according to the summarized record document.

[0042] Compared with the prior art, the beneficial effects of the present invention are:

[0043] 1. This invention collects digital signatures and text information of electronic invoices and binds them with timestamps. At the same time, it integrates historical reimbursement sequences of the same digital signature, and after standardized processing, extracts the writing features such as the stroke logic and structural layout of the signature to accurately construct the invoice to be detected, thereby improving the accuracy of invoice duplication identification and providing a reliable data foundation for subsequent screening.

[0044] 2. This invention filters invoices within a preset interval by timestamp, compares the similarity of writing features in pairs to mark suspected duplicate invoices, and then forms a sequence to be detected by classifying writing features and sorting by timestamp. After comparison with historical reimbursement sequences, it performs differentiated processing, which greatly reduces the amount of invalid manual review and solves the problem of low review efficiency caused by repeated submissions of electronic invoices by multiple departments.

[0045] 3. This invention collects the results of manual review, summarizes the judgment information and related data of the invoices, and updates the reimbursement history of the corresponding digital signature, forming a complete data closed loop, which is conducive to the traceability and verification of subsequent invoice circulation. Attached Figure Description

[0046] Figure 1 This is a flowchart illustrating an intelligent identification method for preventing duplicate receipts according to the present invention.

[0047] Figure 2 This is a schematic diagram of the structure of an intelligent identification system for preventing duplicate receipts according to the present invention. Detailed Implementation

[0048] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0049] Example 1: As Figure 1 As shown, the present invention provides a technical solution, an intelligent identification method for preventing duplicate receipts, the intelligent identification method comprising the following steps:

[0050] Step S1: Collect the digital signature and text information of the electronic invoice and attach a timestamp and historical reimbursement behavior, wherein the historical reimbursement behavior is the reimbursement sequence of all invoices with the same digital signature;

[0051] The specific steps of step S1 are as follows:

[0052] Step S1-1: Collect the digital signature and text information that comes with the electronic ticket, wherein the text information includes all text information on the electronic ticket;

[0053] Step S1-2: Generate a timestamp based on the real-time submission time of the electronic ticket, and bind and associate the timestamp with the digital signature and text information of the corresponding electronic ticket to form a preliminary data set;

[0054] Steps S1-3: Extract all past expense reimbursement data corresponding to a certain digital signature, organize them in the order of reimbursement time to form a complete expense reimbursement sequence, and construct historical reimbursement behavior;

[0055] Step S2: Clean and standardize the collected digital signatures and text information, extract the writing features of the digital signatures, and construct the invoice to be tested;

[0056] The specific steps of step S2 are as follows:

[0057] Step S2-1: Organize the digital signature according to the system's preset format and convert the text information into the preset format.

[0058] Step S2-2: Extract writing features from the regularized digital signature, including the stroke logic and structural layout of the signature;

[0059] Step S2-3: The extracted writing features, converted text information, timestamps, and historical reimbursement behavior are correlated and integrated to form the invoice to be tested;

[0060] Step S3: Extract all tickets to be detected within a preset range before and after the timestamp, collect all writing features, and mark tickets to be detected with a similarity greater than a preset threshold as suspected duplicate tickets.

[0061] The specific steps of step S3 are as follows:

[0062] Step S3-1: Extract a ticket to be tested. Based on its timestamp and the system's preset timestamp filtering range, construct a preset time interval for the ticket to be tested. From all the tickets to be tested, extract all the tickets whose timestamps fall within the preset time interval to form a temporary set of tickets to be tested.

[0063] Step S3-2: Extract the digital signature writing features of each document to be tested from the temporary set of documents to be tested, and form a temporary writing feature set from all the extracted writing features.

[0064] Step S3-3: Using a preset feature similarity calculation method, perform pairwise comparison calculations on all writing features in the temporary writing feature set to obtain the similarity value between each pair of writing features. Compare the calculated similarity value with the system's preset threshold. If the similarity value of two sets of writing features is greater than the system's preset threshold, then mark the tickets to be detected associated with these two sets of features as suspected duplicate tickets. Summarize and organize all the information marked as suspected duplicate tickets to form a list of suspected duplicate tickets.

[0065] Step S4: Extract all suspected duplicate identification tickets, classify them according to writing characteristics, sort the suspected duplicate identification tickets according to timestamps to form a detection sequence, query whether there is a ticket reimbursement sequence with the same detection sequence in the historical reimbursement behavior. If there is, mark it as passed. If not, reimburse the first suspected duplicate identification ticket and transfer the subsequent suspected duplicate identification tickets to manual review.

[0066] The specific steps of step S4 are as follows:

[0067] Step S4-1: Classify all suspected duplicate identification tickets according to the extracted writing features, and group suspected duplicate identification tickets with completely identical writing features into the same category;

[0068] Step S4-2: For suspected duplicate identification tickets within each category, sort them according to their bound timestamp information, and associate and bind the digital signature and standardized text information of each suspected duplicate identification ticket with the sequence to be detected to form the sequence to be detected for each category.

[0069] Step S4-3: Extract the invoice reimbursement sequence from the historical reimbursement behavior associated with the digital signature corresponding to the sequence to be detected. Compare the sequence to be detected with the extracted historical reimbursement sequence. If the comparison result shows that there is a completely identical invoice reimbursement sequence, then all suspected duplicate invoices in this category are marked as passed. If the comparison result shows that there is no completely identical invoice reimbursement sequence, then the suspected duplicate invoice ranked first will be automatically reimbursed, and the remaining invoices will be transferred to the manual review process. Record the comparison result and the corresponding invoice association information.

[0070] Step S5: Collect the results of manual review and update the corresponding invoice reimbursement history.

[0071] The specific steps of step S5 are as follows:

[0072] Step S5-1: Collect the comparison results generated by the manual review process, and summarize the comparison results with the digital signature, standardized text information and timestamp already bound to the invoice to form a record document;

[0073] Step S5-2: Update the expense reimbursement history of the invoice corresponding to the digital signature according to the record document.

[0074] For example:

[0075] Collect electronic invoices submitted by a business unit, extract the unique digital signature of the invoice, and all text information on the invoice, including core information such as invoice type, transaction counterparty, transaction content, amount-related description, and invoice date;

[0076] Based on the real-time time of the submission of the invoice to the organization's reimbursement system, a unique timestamp is generated. This timestamp is then bound to a digital signature and text information to form a preliminary data set containing the core elements of the invoice. Subsequently, all past reimbursement data corresponding to the aforementioned digital signature are extracted and organized into a complete invoice reimbursement sequence according to the order of submission. This constructs the entity's historical reimbursement behavior, providing a universal historical reference for subsequent duplicate identification.

[0077] The digital signatures are standardized according to the pre-set format requirements of the organization's expense reimbursement system. Simultaneously, the text information of the receipts is converted into the system's pre-set unified encoding format to eliminate recognition errors caused by differences in filling habits and data formats among various business units. Next, common writing features are extracted from the standardized digital signatures, including the stroke logic and structural layout of the signature.

[0078] Finally, the extracted writing features and converted text information are linked and integrated with the previously bound timestamps and constructed historical reimbursement behaviors to form a list of invoices to be detected that can be used for subsequent duplicate detection.

[0079] Using the timestamp of a specific invoice to be tested as a baseline, and combining it with the system's preset time range of "several regular business cycles before and after," a preset time interval is determined. From all invoices to be tested within the organization's reimbursement system, all invoices whose timestamps fall within this preset time interval are selected to form a temporary invoice set to be tested. This step eliminates irrelevant invoices from different periods, reduces invalid comparisons, and improves recognition efficiency. Next, using the system's preset feature similarity calculation method, pairwise comparisons are performed on the handwriting features of all invoices in the temporary invoice set to be tested, obtaining the similarity value between each pair of handwriting features. The calculation results are compared with a system preset threshold. If the similarity between two pairs of handwriting features is greater than the threshold, both invoices are jointly marked as suspected duplicate invoices. Finally, the information of all invoices marked as suspected duplicate invoices is summarized to form a list of suspected duplicate invoices.

[0080] Based on the extracted writing characteristics, all invoices in the suspected duplicate invoice list are categorized, with invoices exhibiting completely identical writing characteristics grouped into the same category—for example, the same service consumption invoice submitted by the same submitting entity in two different business units is grouped into one category. Within each category, suspected duplicate invoices are sorted according to their associated timestamp information to determine the order of submission. The digital signature and standardized text information of each invoice are then associated with the sorting result to form a detection sequence for each category. Subsequently, the reimbursement sequence of invoices associated with the digital signatures of the detection sequence is extracted from historical reimbursement activities. The detection sequence is compared with the historical reimbursement sequence: if a completely identical reimbursement sequence exists, all suspected duplicate invoices in that category are marked as "passed"; if no completely identical reimbursement sequence exists, the suspected duplicate invoice ranked first undergoes an automatic reimbursement process, while the remaining invoices are transferred to a manual review process. The comparison results and the associated information of the invoices are recorded in detail.

[0081] The comparison results generated from the manual review process are collected. For some invoices transferred for review, those confirmed as "compliant filings required for cross-business unit collaboration" are judged as "passed." These comparison results are then summarized along with the corresponding invoice's bound digital signature, standardized text information, and timestamp to form a complete record document. This ensures that the judgment process for each invoice is traceable and complies with the organization's reimbursement compliance management requirements. Finally, based on the information in the record document, the historical reimbursement records corresponding to the invoice's digital signature are updated. The comparison results of this reimbursement and the business scenario information associated with the invoice are added to the historical sequence, providing more business-relevant historical data support for the repeated identification of similar cross-business unit invoices in the future.

[0082] Example 2, as Figure 2 As shown, the present invention provides an intelligent identification system for preventing duplicate receipts. The intelligent identification system includes a data acquisition and construction module, a data feature extraction module, a similarity judgment module, a classification and comparison processing module, and a result record update module.

[0083] The data acquisition and construction module is used to collect the digital signatures and text information of electronic invoices and generate timestamps, while constructing a historical reimbursement sequence for all invoices corresponding to the same digital signature. The data feature extraction module is used to clean and standardize the collected digital signatures and text information, extract the writing features of the digital signatures, and associate them with timestamps and historical reimbursement behaviors to construct invoices to be detected. The similarity judgment module is used to filter invoices within a preset time interval based on the timestamps of the invoices to be detected, and to mark suspected duplicate invoices by calculating the similarity of writing features. The classification and comparison processing module is used to classify suspected duplicate invoices according to writing features, sort them according to timestamps to form a sequence to be detected, and perform corresponding processing after comparing them with historical reimbursement sequences. The result recording and updating module is used to collect the comparison results generated by the manual review process, summarize the results with the invoice information, and update the historical reimbursement records of the corresponding digital signatures.

[0084] The output of the data acquisition and construction module is electrically connected to the input of the data feature extraction module; the output of the data feature extraction module is electrically connected to the input of the screening similarity determination module; the output of the screening similarity determination module is electrically connected to the input of the classification comparison processing module; and the output of the classification comparison processing module is electrically connected to the input of the result recording and updating module.

[0085] The data acquisition and construction module includes an information acquisition unit and a historical behavior construction unit;

[0086] The information collection unit is used to collect the digital signature of the electronic invoice and the text information consisting of all the text on the invoice, and generate a timestamp associated with the invoice based on the real-time submission time of the invoice; the historical behavior construction unit is used to extract all past invoice reimbursement data corresponding to a certain digital signature, and organize them into a complete invoice reimbursement sequence according to the reimbursement time.

[0087] The data feature extraction module includes a data regularization unit and a feature extraction unit;

[0088] The data normalization unit is used to normalize the collected digital signatures according to the system's preset format, and at the same time convert the text information into the system's preset format; the feature extraction unit is used to extract the stroke logic and structural layout writing features from the normalized digital signatures, and integrate them with the converted text information, timestamps and historical reimbursement behavior to form the invoice to be tested.

[0089] The similarity determination module includes a time-based filtering unit and a similarity determination unit;

[0090] The time filtering unit is used to construct a preset time interval based on the timestamp of a certain ticket to be detected and the preset timestamp filtering range of the system, and then extract tickets whose timestamps fall within the interval from all tickets to be detected to form a temporary ticket set to be detected; the similarity determination unit is used to perform pairwise comparison calculation of the writing features of all tickets in the temporary ticket set to be detected using a preset feature similarity calculation method, and after comparing the similarity value with a preset threshold, mark suspected duplicate identification tickets and summarize them.

[0091] The classification and comparison processing module includes a ticket classification and sorting unit and a comparison result processing unit;

[0092] The invoice classification and sorting unit is used to classify all suspected duplicate invoices according to the extracted writing features, group invoices with the same writing features into the same category, and then sort the invoices in the category according to the timestamp and associate information to form a sequence to be detected; the comparison result processing unit is used to extract the historical reimbursement sequence corresponding to the digital signature of the sequence to be detected, compare the two, if they match, mark the invoices in the category as passed, if they do not match, the first invoice is automatically reimbursed, and the rest are transferred to manual review and the results are recorded.

[0093] The result recording and updating module includes a result collection unit and a record updating unit;

[0094] The result collection unit is used to collect the invoice comparison results generated by the manual review process, and summarize the results with the digital signature, standardized text information and timestamp bound to the invoice to form a record document; the record update unit is used to update the invoice reimbursement history of the digital signature corresponding to the invoice according to the summarized record document.

[0095] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. An intelligent identification method for preventing duplicate receipts, characterized in that: The intelligent recognition method includes the following steps: Step S1: Collect the digital signature and text information of the electronic invoice and attach a timestamp and historical reimbursement behavior, wherein the historical reimbursement behavior is the reimbursement sequence of all invoices with the same digital signature; The specific steps of step S1 are as follows: Step S1-1: Collect the digital signature and text information that comes with the electronic ticket, wherein the text information includes all text information on the electronic ticket; Step S1-2: Generate a timestamp based on the real-time submission time of the electronic ticket, and bind and associate the timestamp with the digital signature and text information of the corresponding electronic ticket to form a preliminary data set; Steps S1-3: Extract all past expense reimbursement data corresponding to a certain digital signature, organize them in the order of reimbursement time to form a complete expense reimbursement sequence, and construct historical reimbursement behavior; Step S2: Clean and standardize the collected digital signatures and text information, extract the writing features of the digital signatures, and construct the invoice to be tested; Step S3: Extract all tickets to be detected within a preset range before and after the timestamp, collect all writing features, and mark tickets to be detected with a similarity greater than a preset threshold as suspected duplicate tickets. Step S4: Extract all suspected duplicate identification tickets, classify them according to writing characteristics, sort the suspected duplicate identification tickets according to timestamps to form a detection sequence, query whether there is a ticket reimbursement sequence with the same detection sequence in the historical reimbursement behavior. If there is, mark it as passed. If not, reimburse the first suspected duplicate identification ticket and transfer the subsequent suspected duplicate identification tickets to manual review. Step S5: Collect the results of manual review and update the corresponding invoice reimbursement history.

2. The intelligent identification method for preventing duplicate receipts according to claim 1, characterized in that: The specific steps of step S2 are as follows: Step S2-1: Organize the digital signature according to the system's preset format and convert the text information into the preset format. Step S2-2: Extract writing features from the regularized digital signature, including the stroke logic and structural layout of the signature; Step S2-3: The extracted writing features, converted text information, timestamps, and historical reimbursement behavior are associated and integrated to form the invoice to be tested.

3. The intelligent identification method for preventing duplicate receipts according to claim 2, characterized in that: The specific steps of step S3 are as follows: Step S3-1: Extract a ticket to be tested. Based on its timestamp and the system's preset timestamp filtering range, construct a preset time interval for the ticket to be tested. From all the tickets to be tested, extract all the tickets whose timestamps fall within the preset time interval to form a temporary set of tickets to be tested. Step S3-2: Extract the digital signature writing features of each document to be tested from the temporary set of documents to be tested, and form a temporary writing feature set from all the extracted writing features. Step S3-3: Using a preset feature similarity calculation method, perform pairwise comparisons of all writing features in the temporary writing feature set to obtain the similarity value between each pair of writing features. Compare the calculated similarity value with the system's preset threshold. If the similarity value of two sets of writing features is greater than the system's preset threshold, then the tickets to be detected associated with these two sets of features are jointly marked as suspected duplicate tickets. All information marked as suspected duplicate tickets is summarized and organized to form a list of suspected duplicate tickets.

4. The intelligent identification method for preventing duplicate receipts according to claim 3, characterized in that: The specific steps of step S4 are as follows: Step S4-1: Classify all suspected duplicate identification tickets according to the extracted writing features, and group suspected duplicate identification tickets with completely identical writing features into the same category; Step S4-2: For suspected duplicate identification tickets within each category, sort them according to their bound timestamp information, and associate and bind the digital signature and standardized text information of each suspected duplicate identification ticket with the sequence to be detected to form the sequence to be detected for each category. Step S4-3: Extract the invoice reimbursement sequence from the historical reimbursement behavior associated with the digital signature corresponding to the sequence to be detected. Compare the sequence to be detected with the extracted historical reimbursement sequence. If the comparison result shows that there is a completely identical invoice reimbursement sequence, then all suspected duplicate invoices in this category are marked as passed. If the comparison result shows that there is no completely identical invoice reimbursement sequence, then the suspected duplicate invoice ranked first will be automatically reimbursed, and the remaining invoices will be transferred to the manual review process. Record the comparison result and the corresponding invoice association information.

5. The intelligent identification method for preventing duplicate receipts according to claim 4, characterized in that: The specific steps of step S5 are as follows: Step S5-1: Collect the comparison results generated by the manual review process, and summarize the comparison results with the digital signature, standardized text information and timestamp already bound to the invoice to form a record document; Step S5-2: Update the expense reimbursement history of the invoice corresponding to the digital signature according to the record document.

6. An intelligent identification system for preventing duplicate receipts, which is applied to the intelligent identification method for preventing duplicate receipts as described in any one of claims 1-5, characterized in that: The intelligent recognition system includes a data acquisition and construction module, a data feature extraction module, a similarity judgment module, a classification and comparison processing module, and a result recording and updating module. The data acquisition and construction module is used to collect the digital signatures and text information of electronic invoices and generate timestamps, while constructing a historical reimbursement sequence for all invoices corresponding to the same digital signature. The data feature extraction module is used to clean and standardize the collected digital signatures and text information, extract the writing features of the digital signatures, and associate them with timestamps and historical reimbursement behaviors to construct invoices to be detected. The similarity judgment module is used to filter invoices within a preset time interval based on the timestamps of the invoices to be detected, and to mark suspected duplicate invoices by calculating the similarity of writing features. The classification and comparison processing module is used to classify suspected duplicate invoices according to writing features, sort them according to timestamps to form a sequence to be detected, and perform corresponding processing after comparing them with historical reimbursement sequences. The result recording and updating module is used to collect the comparison results generated by the manual review process, summarize the results with the invoice information, and update the historical reimbursement records of the corresponding digital signatures. The output of the data acquisition and construction module is electrically connected to the input of the data feature extraction module; the output of the data feature extraction module is electrically connected to the input of the screening similarity determination module; the output of the screening similarity determination module is electrically connected to the input of the classification comparison processing module; and the output of the classification comparison processing module is electrically connected to the input of the result recording and updating module.

7. The intelligent identification system for preventing duplicate receipts according to claim 6, characterized in that: The data acquisition and construction module includes an information acquisition unit and a historical behavior construction unit; The information collection unit is used to collect the digital signature of the electronic invoice and the text information consisting of all the text on the invoice, and generate a timestamp associated with the invoice based on the real-time submission time of the invoice; the historical behavior construction unit is used to extract all past invoice reimbursement data corresponding to a certain digital signature, and organize them into a complete invoice reimbursement sequence according to the reimbursement time.

8. The intelligent identification system for preventing duplicate receipts according to claim 6, characterized in that: The data feature extraction module includes a data regularization unit and a feature extraction unit; The data normalization unit is used to normalize the collected digital signatures according to the system's preset format, and at the same time convert the text information into the system's preset format; the feature extraction unit is used to extract the stroke logic and structural layout writing features from the normalized digital signatures, and integrate them with the converted text information, timestamps and historical reimbursement behavior to form the invoice to be tested. The similarity determination module includes a time-based filtering unit and a similarity determination unit; The time filtering unit is used to construct a preset time interval based on the timestamp of a certain ticket to be detected and the preset timestamp filtering range of the system, and then extract tickets whose timestamps fall within the interval from all tickets to be detected to form a temporary ticket set to be detected; the similarity determination unit is used to perform pairwise comparison calculation of the writing features of all tickets in the temporary ticket set to be detected using a preset feature similarity calculation method, and after comparing the similarity value with a preset threshold, mark suspected duplicate identification tickets and summarize them.

9. The intelligent identification system for preventing duplicate receipts according to claim 6, characterized in that: The classification and comparison processing module includes a ticket classification and sorting unit and a comparison result processing unit; The invoice classification and sorting unit is used to classify all suspected duplicate invoices according to the extracted writing features, group invoices with the same writing features into the same category, and then sort the invoices in the category according to the timestamp and associate information to form a sequence to be detected; the comparison result processing unit is used to extract the historical reimbursement sequence corresponding to the digital signature of the sequence to be detected, compare the two, if they match, mark the invoices in the category as passed, if they do not match, the first invoice is automatically reimbursed, and the rest are transferred to manual review and the results are recorded. The result recording and updating module includes a result collection unit and a record updating unit; The result collection unit is used to collect the invoice comparison results generated by the manual review process, and summarize the results with the digital signature, standardized text information and timestamp bound to the invoice to form a record document; the record update unit is used to update the invoice reimbursement history of the digital signature corresponding to the invoice according to the summarized record document.

Images