Bill auditing method and device, computer equipment, storage medium and computer program product
By determining the target bill field of the bill to be reviewed in RPA technology and relocating the review rules, the problem of inefficient writing review rules is solved, and efficient review of bills in new fields is achieved.
Patent Information
- Application Number
- CN202510221995.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-27
- Publication Date
- 2025-06-27
AI Technical Summary
When using RPA technology for intelligent auditing, it is less efficient to write audit rules, especially when encountering bills to be reviewed in new fields, technicians need to manually write rules, resulting in inefficient audits.
By determining the target bill field of the bill to be reviewed and finding similar fields in the preset bill field as a reference, alternative review rules are generated based on the similarity between the review point information of the bill to be reviewed and the review point information of the reference area, thereby migrating the review rules to the new domain.
There is no need for technicians to manually write rules for each audit field, which improves the efficiency of generating audit rules and improves the efficiency of auditing bills.
Smart Images
Figure CN120218838A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of artificial intelligence, and particularly to a bill auditing method, apparatus, computer device, storage medium, and computer program product. Background Art
[0002] Today, as enterprises continuously pursue efficiency improvement and cost reduction, RPA (Robotic Process Automation) technology is gradually becoming the focus and is widely applied in various fields. By simulating human operations on a computer, RPA technology realizes the automation of business processes, thus significantly improving work efficiency and reducing labor costs.
[0003] When applied to the field of intelligent auditing, since the RPA service requires technicians to write auditing rules for it before it can run normally, technicians need to customize auditing rules for each auditing field before using the RPA service for automated operations. Among them, the auditing rules are completely written relying on the experience of technicians. As a result, when there are bills to be audited belonging to a new field, the efficiency of writing the corresponding auditing rules for this field is low, and further, the efficiency of auditing the bills to be audited is also low. Summary of the Invention
[0004] Based on this, in view of the above technical problems, it is necessary to provide a bill auditing method, apparatus, computer device, storage medium, and computer program product.
[0005] In a first aspect, the present application provides a bill auditing method. The method includes:
[0006] Determine the target bill field corresponding to the bill to be audited. In the case where there is no corresponding auditing rule for the target bill field, determine the reference bill field corresponding to the target bill field among each preset bill field;
[0007] Based on the bill information included in the bill to be audited, determine each alternative auditing point corresponding to the bill to be audited, and respectively determine the auditing point information corresponding to each alternative auditing point. The auditing point information includes semantic information and data type information;
[0008] For any one of the alternative auditing points, according to the similarity between the auditing point information corresponding to the alternative auditing point and the auditing point information corresponding to each auditing point in the reference bill field, determine a reference auditing point from each auditing point, and generate an alternative auditing rule corresponding to the alternative auditing point based on the auditing rule corresponding to the reference auditing point;
[0009] Determine the target audit rule corresponding to the target bill area based on each of the alternative audit rules, and audit the bill to be audited based on the target audit rule.
[0010] In one embodiment, the determining the reference audit point from each of the audit points according to the similarity between the audit point information corresponding to the alternative audit point and the audit point information corresponding to each audit point in the reference bill area includes:
[0011] Determine the semantic similarity between the semantic information of the alternative audit point and the semantic information of each of the audit points respectively, and determine the first reference audit point from each of the audit points according to the semantic similarity corresponding to each of the audit points;
[0012] In the case where there is a target first reference audit point in each of the first reference audit points whose data type information is the same as the data type information of the alternative audit point, use the target first reference audit point as the reference audit point.
[0013] In one embodiment, the alternative audit rule includes information to be filled in, and the determining the target audit rule corresponding to the target bill area based on each of the alternative audit rules includes:
[0014] Display each of the alternative audit points, and in response to a selection operation on a target alternative audit point among each of the alternative audit points, display each of the alternative audit rules corresponding to the target alternative audit point;
[0015] In response to a selection operation on a target alternative audit rule among each of the alternative audit rules, display the information to be filled in corresponding to the target alternative audit rule, and in response to an editing operation on the information to be filled in, obtain the target audit rule corresponding to the target bill area.
[0016] In one embodiment, the audit rule includes rule framework information and rule customization information, and the generating the alternative audit rule corresponding to the alternative audit point based on the audit rule corresponding to the reference audit point includes:
[0017] Use the rule framework information in the audit rule corresponding to the reference audit point as the rule framework information corresponding to the alternative audit point;
[0018] Determine the information to be filled in and the prompt information corresponding to the alternative audit point according to the rule customization information in the audit rule corresponding to the reference audit point, where the prompt information is used to prompt filling in the information to be filled in based on the rule customization information;
[0019] Generate the alternative audit rule according to the rule framework information, the information to be filled in, and the prompt information.
[0020] In one embodiment, the audit rule corresponds to a rule type, and the rule type includes a numerical range rule type, a text matching rule type, and a text content rule type;
[0021] When the rule type is the numerical range rule type, the rule customization information is a numerical range;
[0022] When the rule type is the text matching rule type, the rule customization information is the text to be matched;
[0023] When the rule type is the text content rule type, the rule customization information is text semantic information.
[0024] In one embodiment, in the case where the target first reference audit point does not exist, the method further includes:
[0025] Obtain a basic audit rule that matches the data type information of the alternative audit point, and generate an alternative audit rule corresponding to the alternative audit point according to the basic audit rule.
[0026] In a second aspect, the present application also provides a bill audit device. The device includes:
[0027] A first determination module, configured to determine the target bill field corresponding to the bill to be audited, and determine the reference bill field corresponding to the target bill field in each preset bill field when there is no corresponding audit rule in the target bill field;
[0028] A second determination module, configured to determine each alternative audit point corresponding to the bill to be audited based on the bill information included in the bill to be audited, and respectively determine the audit point information corresponding to each alternative audit point, where the audit point information includes semantic information and data type information;
[0029] A generation module, configured to, for any one of the alternative audit points, determine a reference audit point from each of the audit points according to the similarity between the audit point information corresponding to the alternative audit point and the audit point information corresponding to each audit point in the reference bill field, and generate an alternative audit rule corresponding to the alternative audit point based on the audit rule corresponding to the reference audit point;
[0030] A third determination module, configured to determine the target audit rule corresponding to the target bill field based on each of the alternative audit rules, and audit the bill to be audited based on the target audit rule.
[0031] In one embodiment, the generation module is further configured to:
[0032] Determine the semantic similarity between the semantic information of the alternative review points and the semantic information of each review point respectively, and determine the first reference review point from each review point according to the semantic similarity corresponding to each review point;
[0033] In the case where there is a target first reference review point with the same data type information as the data type information of the alternative review point among the first reference review points, use the target first reference review point as the reference review point.
[0034] In one embodiment, the alternative review rules include information to be filled in, and the third determination module is further configured to:
[0035] Display each alternative review point, and in response to a selection operation on a target alternative review point among each alternative review point, display each alternative review rule corresponding to the target alternative review point;
[0036] In response to a selection operation on a target alternative review rule among each alternative review rule, display the information to be filled in corresponding to the target alternative review rule, and in response to an editing operation on the information to be filled in, obtain the target review rule corresponding to the target bill field.
[0037] In one embodiment, the review rule includes rule framework information and rule customization information, and the generation module is further configured to:
[0038] Use the rule framework information in the review rule corresponding to the reference review point as the rule framework information corresponding to the alternative review point;
[0039] Determine the information to be filled in and the prompt information corresponding to the alternative review point according to the rule customization information in the review rule corresponding to the reference review point, where the prompt information is used to prompt filling in the information to be filled in based on the rule customization information;
[0040] Generate the alternative review rule according to the rule framework information, the information to be filled in, and the prompt information.
[0041] In one embodiment, the review rule corresponds to a rule type, and the rule type includes a numerical range rule type, a text matching rule type, and a text content rule type;
[0042] When the rule type is the numerical range rule type, the rule customization information is a numerical range;
[0043] When the rule type is the text matching rule type, the rule customization information is the text to be matched;
[0044] When the rule type is the text content rule type, the rule customization information is text semantic information.
[0045] In one embodiment, when the target first reference review point does not exist, the generating module is further configured to:
[0046] Obtain a basic review rule that matches the data type information of the alternative review point, and generate an alternative review rule corresponding to the alternative review point according to the basic review rule.
[0047] In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the above-mentioned method is implemented.
[0048] In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned method is implemented.
[0049] In a fifth aspect, the present application further provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the above-mentioned method is implemented.
[0050] When there is no corresponding review rule in the target bill field to which the bill to be reviewed belongs, the above-mentioned bill review method, device, computer device, storage medium, and computer program product determine the reference bill field corresponding to the target bill field, and based on the review point information of each alternative review point of the bill to be reviewed, select a reference review point from the review points corresponding to the reference bill field, and then migrate the review rule corresponding to the reference review point to the alternative review point to obtain the target review rule corresponding to the target bill field. The embodiments of the present application can migrate the review rules in other fields to a new field when a new bill review field appears, without the need for technicians to manually write review rules for each review field, improving the generation efficiency of review rules, and thus improving the review efficiency of the bill to be reviewed. BRIEF DESCRIPTION OF THE DRAWINGS
[0051] Figure 1 It is a schematic flowchart of a bill review method in an embodiment;
[0052] Figure 2 It is a schematic flowchart of determining a reference review point in step 106 in an embodiment;
[0053] Figure 3 It is a schematic flowchart of step 108 in an embodiment;
[0054] Figure 4 It is a schematic flowchart of generating alternative audit rules in step 106 in an embodiment;
[0055] Figure 5 It is a structural block diagram of a bill audit device in an embodiment;
[0056] Figure 6 It is an internal structure diagram of a computer device in an embodiment. Detailed implementation manners
[0057] In order to make the objectives, technical solutions and advantages of the present application more clear and understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0058] In one embodiment, as Figure 1 shown, a bill audit method is provided. In this embodiment, an example is given where this method is applied to a server. It can be understood that this method can also be applied to a terminal, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps:
[0059] Step 102, determine the target bill field corresponding to the bill to be audited. In the case that there is no corresponding audit rule in the target bill field, determine the reference bill field corresponding to the target bill field in each preset bill field.
[0060] In the embodiments of the present application, the bill to be audited can be any bill such as an invoice, a ticket, a travel itinerary receipt, etc. The target bill field refers to the field to which the expense type corresponding to the bill to be audited belongs. For example, when the bill to be audited is a train ticket, the target bill field can be the railway field, and when the bill to be audited is an air travel itinerary receipt, the target bill field can be the aviation field.
[0061] The bill information on the bill to be audited can be semantically recognized by using a large language model or other trained semantic recognition models to obtain the target bill field corresponding to the bill to be audited. Further, the server can also store all the bill fields that have been recognized for each bill to be audited in the past. After semantically recognizing a new bill to be audited and obtaining the semantic vector of the new bill to be audited, a classification model can also be used to determine the confidence of this semantic vector belonging to each bill field. If there is a bill field with the highest confidence and greater than the threshold, it is determined that the target bill field of the new bill to be audited is this bill field. If there is no bill field with a confidence greater than the threshold, a new target bill field can be created for the bill to be audited based on the semantic vector, so as to classify the bill to be audited into the bill fields with existing audit rules as much as possible and reduce the frequency of newly constructing audit rules for the bill to be audited.
[0062] After obtaining the target bill area, it can be determined whether the audit rules corresponding to the target bill area are stored in the server. The audit rules are used to judge whether the bill information recorded on the bill to be audited meets the requirements. For example, whether the amount on the bill to be audited is within the preset range, whether the date corresponding to the bill to be audited is consistent with that filled in the expense reimbursement application corresponding to the bill to be audited, and so on. If such audit rules exist in the server, the audit rules are used to audit the bill to be audited. If such audit rules do not exist in the server, it is necessary to newly construct audit rules for the target bill area. These audit rules can be migrated from a reference bill area that is relatively similar to the target bill area. The reference bill area is one of the preset bill areas. The preset bill areas can be set in advance by technical personnel, and the technical personnel can select the bill areas with relatively perfect and representative audit rules as the preset bill areas.
[0063] The semantic similarity between the target bill area and each preset bill area can be calculated, and one or more preset bill areas with the highest semantic similarity can be selected as the reference bill areas. Further, key audit points can also be determined in advance for each preset bill area. The key audit points are one or more of the audit points corresponding to the preset bill area. The audit points are the bill information that needs to be audited on the bill to be audited in the preset bill area, such as amount, date, purchaser information, etc. The key audit points are also the audit points that can better represent the characteristics of the preset bill area. For example, for the railway field, the key audit points can be itinerary information, seat class, etc., and for the insurance field, the key audit points can be insured subject, insurance period, etc. Which specific audit points are the key audit points can be set by technical personnel according to actual needs.
[0064] The server can identify the semantic information corresponding to each bill information on the bill to be audited through a large language model or other trained semantic recognition models, and then calculate the similarity between each key audit point in each preset bill area and the semantic information corresponding to each bill information respectively. If there is semantic information with a similarity greater than the threshold, it can be considered that the corresponding bill information exists for the key audit point on the bill to be audited. If, for a certain preset bill area, the number or proportion of key audit points with corresponding bill information exceeds the threshold, it can be considered that the audit rules of this preset bill area have a high probability of being migrated to the target bill area, and this preset bill area can be used as the reference bill area.
[0065] Step 104: Based on the bill information included in the bill to be audited, determine each alternative audit point corresponding to the bill to be audited, and respectively determine the audit point information corresponding to each alternative audit point. The audit point information includes semantic information and data type information.
[0066] In the embodiments of the present application, the server can identify the semantic information corresponding to each bill information on the bill to be audited through a large language model or other trained semantic recognition models, and then determine the parts in the bill information that may need to be audited based on the semantic information, that is, the alternative audit points corresponding to the bill to be audited. Taking the bill to be audited as a train ticket as an example, the semantic information corresponding to the bill information may include invoice number, invoice date, itinerary information, seat class information, fare, passenger ID information, purchaser name, etc. Among them, the parts that may need to be audited may include itinerary information, seat class information, fare, and purchaser name. Therefore, the alternative audit points determined based on the semantic information can be the above four pieces of information. The alternative audit points can be obtained by constructing prompt words based on the semantic information and then inputting the prompt words into the large language model to make the large language model select from each semantic information.
[0067] After determining the alternative audit points, the audit point information corresponding to the alternative audit points can be generated. The audit point information includes semantic information and data type information. The semantic information is the meaning of the bill information corresponding to the alternative audit point, and the data type information is the data type of the bill information corresponding to the alternative audit point, such as text, number, image, time, etc. Taking the alternative audit point "itinerary information" as an example, the semantic information of this alternative audit point can be "departure and arrival locations of transportation", and the data type information can be text. The semantic information can be the semantic information extracted when determining the alternative audit points above. The data type information can be identified and determined through the large language model. Other information can also be included in the audit point information, such as the importance priority of this audit point, the relationship information with other audit points (for example, there is a corresponding relationship between the purchaser name and the purchaser's unified social credit code), etc. The embodiments of the present application do not make specific limitations.
[0068] Step 106, for any alternative audit point, determine a reference audit point from each audit point according to the similarity between the audit point information corresponding to the alternative audit point and the audit point information corresponding to each audit point in the reference bill field, and generate an alternative audit rule corresponding to the alternative audit point based on the audit rule corresponding to the reference audit point.
[0069] In the embodiments of the present application, for each alternative audit point, it can be determined whether there is a reference audit point similar to the alternative audit point in the reference bill field. If there is, the audit rule of the reference audit point can be migrated to the alternative audit point to obtain the alternative audit rule corresponding to the alternative audit point.
[0070] It can be determined whether an audit point is similar to an alternative audit point based on the similarity between the audit point information of the alternative audit point and the audit point information of the audit point. For example, the semantic similarity between the semantic information of the alternative audit point and the semantic information of the audit point can be calculated. If the similarity is greater than the threshold, it is determined that the audit point is a reference audit point of the alternative audit point.
[0071] In one embodiment, as Figure 2 shown, in step 106, according to the similarity between the checkpoint information corresponding to the alternative checkpoint and the checkpoint information corresponding to each checkpoint in the reference bill field, the reference checkpoint is determined from each checkpoint, including:
[0072] Step 202, respectively determine the semantic information of the alternative checkpoint and the semantic similarity between the semantic information of each checkpoint, and determine the first reference checkpoint from each checkpoint according to the semantic similarity corresponding to each checkpoint;
[0073] Step 204, in the case that there is a target first reference checkpoint with the same data type information as the alternative checkpoint among the first reference checkpoints, use the target first reference checkpoint as the reference checkpoint.
[0074] In the embodiment of the present application, first, the semantic similarity between the alternative checkpoint and the checkpoint is matched. The first reference checkpoint is the checkpoint whose semantic similarity meets the requirements. For example, a semantic similarity threshold can be set. When there is a checkpoint with a semantic similarity greater than the semantic similarity threshold, use the checkpoint as the first reference checkpoint. Or it can also be the checkpoint with a semantic similarity greater than the semantic similarity threshold and the highest semantic similarity among all checkpoints as the first reference checkpoint. The embodiment of the present application does not make specific limitations.
[0075] After determining the first reference checkpoint, the server further determines whether the data type of the first reference checkpoint is the same as the data type of the alternative checkpoint. Since the audit rules are generally written for specific data types, the audit rules of the first reference checkpoint can be migrated to the alternative checkpoint only when the data types of the first reference checkpoint and the alternative checkpoint are the same. At the same time, since the semantic information is automatically generated by the model, in some special cases, there may be a situation where the semantic information of the alternative checkpoint is similar to the semantic information of a certain checkpoint, but there are substantial differences between these two checkpoints. For example, the semantic information corresponding to the passenger name and the semantic information corresponding to the passenger ID number may be determined to be similar. Verifying whether the data types of the first reference checkpoint and the alternative checkpoint are the same can avoid the situation where the audit rules of substantially different checkpoints are migrated to the alternative checkpoint.
[0076] If there is no first reference review point with semantic similarity meeting the requirements, or there is no target first reference review point with the same data type, the server can generate an alarm message to inform the technician that the review rules in other fields cannot be migrated to the alternative review points in the target bill field. The technician needs to manually write the review rules for the alternative review points, or the alternative review points can also be deleted. Alternatively, the server can also store the basic review rules corresponding to each data type, and these basic review rules can be set by the technician according to the actual situation. For example, the basic review rule for the text data type can be whether the text matches the specified text, the basic review rule for the numeric data type can be whether the number is less than a certain preset value, and the basic review rule for the time data type can be whether the time is within the specified time interval, etc. If there is no first reference review point or target first reference review point, the server can skip the step of generating alternative review rules based on the review rules of the reference review points, obtain the basic review rules that match the data type information of the alternative review points, and generate the alternative review rules corresponding to the alternative review points according to the basic review rules.
[0077] After obtaining the basic review rules or the review rules corresponding to the reference review points, the alternative review rules corresponding to the alternative review points can be generated based on these review rules. A review rule can be divided into two parts: rule framework information and rule custom information. The rule framework information is used to define the logical judgment method corresponding to the rule, and the rule custom information is used to define the specific criteria included in the rule. For example, when the review rule is "the ticket price needs to be less than 1000 yuan", the rule framework information can be "the ticket price is less than the preset value", and the rule custom information can be "1000 yuan". When migrating the review rules corresponding to the reference review points, the rule framework information in the review rules can be used as the rule framework information corresponding to the alternative review points, and at the same time, the rule custom information is adjusted to obtain the rule custom information corresponding to the alternative review points.
[0078] The embodiments of the present application do not specifically limit how to adjust the rule customization information. Exemplarily, a custom information data table can be pre-constructed, and the rule customization information, the field corresponding to the rule customization information, and the review points corresponding to the rule customization information in the field are recorded in the custom information data table. For example, when it is necessary to limit that any transportation expense needs to be less than 1000 yuan, the rule customization information of 1000 yuan can be recorded in the custom information data table, the field corresponding to the rule customization information is set to the transportation field, and the review point is set to the expense. When it is necessary to adjust the rule customization information corresponding to the alternative review points, the custom information review table can be queried. When there is rule customization information in the custom information review table whose corresponding field is similar to the target bill field and the review point is similar to the alternative review point, this rule customization information is used as the rule customization information corresponding to the alternative review point. Or the server can also notify the technician to manually fill in the rule customization information. The embodiments of the present application do not specifically limit this.
[0079] It should be noted that one alternative review point may correspond to multiple reference review points, and one reference review point may also correspond to multiple review rules. Therefore, there may be one or more alternative review rules generated for one alternative review point.
[0080] Step 108: Determine the target review rules corresponding to the target bill field based on each alternative review rule, and review the bill to be reviewed based on the target review rules.
[0081] In the embodiments of the present application, the target review rules to be applied can be determined from each alternative review rule. All alternative review rules can be displayed to the technician, and the technician determines which alternative review rules need to be used as the target review rules. Or the server can also select the target review rules according to certain rules.
[0082] For example, since each alternative review point may correspond to multiple alternative review rules simultaneously, one alternative review rule can be selected for each alternative review point as the target review rule, thereby obtaining the target review rules corresponding to the target bill area. The embodiments of the present application do not specifically limit how to select the target review rule for each alternative review point. The alternative review rule with the highest frequency of occurrence among the review rules corresponding to its respective reference review points in the corresponding rule framework information can be selected as the target review rule, which is equivalent to preferentially retaining the relatively general alternative review rules among the respective reference review points. Or, weights can be assigned to each reference review point based on the similarity between each reference review point and the alternative review point, such that the higher the similarity, the higher the weight. Then, the similarity between the rule framework information of each alternative review rule and the rule framework information of the review rules of each reference review point is determined, and the scores for the alternative review rules are obtained by weighted summation of the similarities according to the weights of the reference review points. The alternative review rule with the highest score is selected as the target review rule.
[0083] After obtaining the target review rules, the target review rules can be implemented in the RPA service for bill review, so that when the RPA service conducts bill review according to the target review rules, it can determine that the bill to be reviewed is non-compliant and return the bill review process to the initiator when the bill to be reviewed does not meet any target review rules, or determine that the bill to be reviewed is compliant and pass the bill review process when the bill to be reviewed meets all the target review rules, thereby realizing the automation of bill review.
[0084] Furthermore, the target review rules can be stored in the server as the review rules for the target bill area. In this way, when the server subsequently receives a bill to be reviewed belonging to the target bill area, it can read the target review rules to review the bill to be reviewed, improving the review efficiency of the bill to be reviewed.
[0085] The bill review method provided by the embodiments of the present application determines the corresponding reference bill area for the target bill area when there are no corresponding review rules for the target bill area to which the bill to be reviewed belongs, and selects reference review points from the review points corresponding to the reference bill area based on the review point information of each alternative review point of the bill to be reviewed. Then, the review rules corresponding to the reference review points are migrated to the alternative review points to obtain the target review rules corresponding to the target bill area. The embodiments of the present application can migrate the review rules in other areas to the new area when a new bill review area appears, without the need for technicians to manually write review rules for each review area, improving the generation efficiency of review rules and thus enhancing the review efficiency of the bill to be reviewed.
[0086] In one embodiment, the alternative review rules include information to be filled in, such as Figure 3As shown in the figure, in step 108, determining the target review rule corresponding to the target bill area based on each alternative review rule includes:
[0087] Step 302, display each alternative review point, and in response to a selection operation for a target alternative review point among each alternative review point, display each alternative review rule corresponding to the target alternative review point;
[0088] Step 304, in response to a selection operation for a target alternative review rule among each alternative review rule, display the information to be filled in corresponding to the target alternative review rule, and in response to an editing operation for the information to be filled in, obtain the target review rule corresponding to the target bill area.
[0089] In the embodiment of the present application, the alternative review rule includes information to be filled in, and the information to be filled in is the information that needs to be filled in by technical personnel corresponding to the custom information of the rule. That is, in the embodiment of the present application, technical personnel can manually fill in the custom information of the rule. After generating each alternative review rule, the server can display each alternative review point and each alternative review rule to technical personnel. Technical personnel can select the target alternative review rule from them, and fill in the custom information of the rule in the target alternative review rule to obtain the target review rule.
[0090] The server can first display each alternative review point to technical personnel. Technical personnel can select the target alternative review point that actually needs to be reviewed from them. Each alternative review point can be arranged in descending order of the importance priority included in the review point information, that is, the alternative review point with a higher importance priority is arranged more forward. The importance priority can be generated by a large language model according to the semantic information of each alternative review point.
[0091] After technical personnel select the target alternative review point, the server then displays each alternative review rule corresponding to the target alternative review point. After technical personnel determine one or more target alternative review rules from each alternative review rule, they can edit the information to be filled in each target alternative review rule to obtain the target review rule corresponding to the target bill area.
[0092] In one embodiment, as Figure 4 shown, in step 106, generating the alternative review rule corresponding to the alternative review point based on the review rule corresponding to the reference review point includes:
[0093] Step 402, use the rule framework information in the review rule corresponding to the reference review point as the rule framework information corresponding to the alternative review point;
[0094] Step 404: Determine the information to be filled and the prompt information corresponding to the alternative review points according to the rule customization information in the review rules corresponding to the reference review points. The prompt information is used to prompt the filling of the information to be filled based on the rule customization information.
[0095] Step 406: Generate alternative review rules based on the rule framework information, the information to be filled, and the prompt information.
[0096] In the embodiments of the present application, after using the rule framework information of the reference review point as the rule framework information of the alternative review point and determining the information to be filled in the review rule based on the rule customization information, the prompt information corresponding to the alternative review point can also be generated based on the rule customization information, so that technicians can fill in the information to be filled with reference to the prompt information. The prompt information may include the rule customization information corresponding to each reference review point, the alternative rule customization information generated by the large language model based on the rule customization information corresponding to each reference review point, and so on.
[0097] For example, in the case where the target bill field is the aviation field, the rule framework information is "the ticket price is less than the preset value", and after the information to be filled determined based on the rule customization information is "the preset value", the prompt information "Fill in a reasonable value. For the railway field, this value is 1000 yuan, and for the aviation field, it is recommended that this value be 3000 yuan" can be generated based on the rule customization information "1000 yuan" in the reference review point. Technicians can then fill in the information to be filled with reference to the rule customization information set in the reference bill field and the alternative rule customization information generated by the large language model.
[0098] In one embodiment, the review rule corresponds to a rule type, and the rule type includes a numerical range rule type, a text matching rule type, and a text content rule type; in the case where the rule type is the numerical range rule type, the rule customization information is a numerical range; in the case where the rule type is the text matching rule type, the rule customization information is the text to be matched; in the case where the rule type is the text content rule type, the rule customization information is text semantic information.
[0099] In the embodiments of the present application, the review rules can be divided into a numerical range rule type, a text matching rule type, and a text content rule type. The numerical range rule type corresponds to review points with a data type of number, and is used to limit the numerical range to which the bill information corresponding to the review point should belong. The text matching rule type corresponds to review points with a data type of text or number, and is used to limit the text that the bill information corresponding to the review point should match. The text content rule type corresponds to review points with a data type of text, and is used to limit the requirements that the content of the bill information corresponding to the review point should meet.
[0100] When the audit rule is of the numerical range rule type, the rule customization information can be a numerical range, and the information to be filled in by the technician in the audit rule can be the upper and lower limits of the numerical range.
[0101] When the audit rule is of the text matching rule type, the rule customization information can be the text to be matched. The information to be filled in by the technician in the audit rule can be text information or the source of the text to be matched. For example, when the audit rule is "the seat class needs to be second-class seat", what the technician can fill in the information to be filled in can be "second-class seat" or the storage address of the text "second-class seat" in a certain data table.
[0102] When the audit rule is of the text content rule type, the rule customization information can be text semantic information. The information to be filled in by the technician in the audit rule can be text information or the source of the text information. For example, when the audit rule is "the ticket needs to contain passenger information", what the technician can fill in the information to be filled in can be "passenger information" or the initiator of the bill audit process, so as to verify whether the initiator of the bill audit process matches the passenger recorded in the ticket.
[0103] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are shown in sequence according to the indication of the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same moment, but can be executed at different moments. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or alternately with at least a part of other steps or steps or stages in other steps.
[0104] Based on the same inventive concept, the embodiments of the present application also provide a bill audit device for implementing the above-mentioned bill audit method. The implementation solutions provided by this device to solve problems are similar to the implementation solutions described in the above method. Therefore, the specific limitations in one or more embodiments of the bill audit device provided below can refer to the limitations on the bill audit method in the above text, and will not be repeated here.
[0105] In one embodiment, as Figure 5 shown, a bill audit device 500 is provided, including: a first determination module 502, a second determination module 504, a generation module 506, and a third determination module 508, where:
[0106] The first determination module 502 is configured to determine the target bill area corresponding to the bill to be audited, and determine the reference bill area corresponding to the target bill area in each preset bill area when there is no corresponding audit rule in the target bill area;
[0107] The second determination module 504 is configured to determine each alternative audit point corresponding to the bill to be audited based on the bill information included in the bill to be audited, and respectively determine the audit point information corresponding to each alternative audit point, where the audit point information includes semantic information and data type information;
[0108] The generation module 506 is configured to, for any one of the alternative audit points, determine a reference audit point from each of the audit points according to the similarity between the audit point information corresponding to the alternative audit point and the audit point information corresponding to each audit point in the reference bill area, and generate an alternative audit rule corresponding to the alternative audit point based on the audit rule corresponding to the reference audit point;
[0109] The third determination module 508 is configured to determine the target audit rule corresponding to the target bill area based on each of the alternative audit rules, and audit the bill to be audited based on the target audit rule.
[0110] When there is no corresponding audit rule in the target bill area to which the bill to be audited belongs, the bill audit device provided by the embodiment of the present application determines the reference bill area corresponding to the target bill area, and selects a reference audit point from the audit points corresponding to the reference bill area based on the audit point information of each alternative audit point of the bill to be audited, and then migrates the audit rule corresponding to the reference audit point to the alternative audit point to obtain the target audit rule corresponding to the target bill area. The embodiment of the present application can migrate the audit rules in other areas to a new area when a new bill audit area appears, without the need for technicians to manually write audit rules for each audit area, improving the generation efficiency of audit rules, and thus improving the audit efficiency of the bill to be audited.
[0111] In one embodiment, the generation module 506 is further configured to:
[0112] Respectively determine the semantic similarity between the semantic information of the alternative audit point and the semantic information of each of the audit points, and determine a first reference audit point from each of the audit points according to the semantic similarity corresponding to each of the audit points;
[0113] When there is a target first reference audit point in each of the first reference audit points where the data type information is the same as the data type information of the alternative audit point, use the target first reference audit point as the reference audit point.
[0114] In one embodiment, the alternative review rules include information to be filled in, and the third determination module 508 is further configured to:
[0115] Display each of the alternative review points, and in response to a selection operation for a target alternative review point among each of the alternative review points, display each of the alternative review rules corresponding to the target alternative review point;
[0116] In response to a selection operation for a target alternative review rule among each of the alternative review rules, display the information to be filled in corresponding to the target alternative review rule, and in response to an editing operation for the information to be filled in, obtain the target review rule corresponding to the target bill area.
[0117] In one embodiment, the review rule includes rule framework information and rule customization information, and the generation module 506 is further configured to:
[0118] Use the rule framework information in the review rule corresponding to the reference review point as the rule framework information corresponding to the alternative review point;
[0119] Determine the information to be filled in and the prompt information corresponding to the alternative review point according to the rule customization information in the review rule corresponding to the reference review point, where the prompt information is used to prompt filling in the information to be filled in based on the rule customization information;
[0120] Generate the alternative review rule according to the rule framework information, the information to be filled in, and the prompt information.
[0121] In one embodiment, the review rule corresponds to a rule type, and the rule type includes a numerical range rule type, a text matching rule type, and a text content rule type;
[0122] When the rule type is the numerical range rule type, the rule customization information is a numerical range;
[0123] When the rule type is the text matching rule type, the rule customization information is the text to be matched;
[0124] When the rule type is the text content rule type, the rule customization information is text semantic information.
[0125] In one embodiment, when the target first reference review point does not exist, the generation module 506 is further configured to:
[0126] Obtain a basic review rule that matches the data type information of the alternative review point, and generate an alternative review rule corresponding to the alternative review point according to the basic review rule.
[0127] Each module in the above device can be implemented in whole or in part by software, hardware, and their combination. Each of the above modules can be embedded in the processor of the computer device in hardware form or be independent of it, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to each of the above modules.
[0128] In one embodiment, a computer device is provided. The computer device can be a server, and its internal structure diagram can be as Figure 6 shown. The computer device includes a processor, a memory, and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it implements a bill review method.
[0129] Those skilled in the art can understand that Figure 6 the structure shown in
[0130] is only a block diagram of a part of the structure related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have a different component layout.
[0131] In one embodiment, a computer device is provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, it implements the steps in the above method embodiments.
[0132] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by the processor, it implements the steps in the above method embodiments.
[0133] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.
[0134] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memories. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in the present application can be general-purpose processors, central processors, graphics processors, digital signal processors, programmable logic devices, data processing logics based on quantum computing, etc., without limitation.
[0135] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.
[0136] The above-described embodiments merely represent several implementation manners of the present application. The description is relatively specific and detailed, but it should not be construed as a limitation on the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.
Claims
1. A bill audit method, characterized in that: The method comprises: Determine the target bill field corresponding to the bill to be reviewed, and if the target bill field does not have a corresponding review rule, determine the reference bill field corresponding to the target bill field in each preset bill field; Based on the bill information contained in the bill to be reviewed, determine each candidate review point corresponding to the bill to be reviewed, and respectively determine the review point information corresponding to each candidate review point, wherein the review point information includes semantic information and data type information; For any of the candidate audit points, according to the similarity between the audit point information corresponding to the candidate audit point and the audit point information corresponding to each audit point in the reference bill field, a reference audit point is determined from the audit points, and based on the audit rule corresponding to the reference audit point, an alternative audit rule corresponding to the candidate audit point is generated; The target audit rule corresponding to the target bill field is determined based on each of the candidate audit rules, and the bill to be audited is audited based on the target audit rule.
2. The method according to claim 1, characterized in that The determining of the reference audit point from each of the audit points according to the similarity between the audit point information corresponding to the candidate audit point and the audit point information corresponding to each of the audit points in the reference bill field comprises: respectively determining the semantic information of the candidate audit points and the semantic similarity between the semantic information of each audit point, and determining a first reference audit point from each audit point according to the semantic similarity corresponding to each audit point; If there is a target first reference audit point in each of the first reference audit points whose data type information is the same as that of the candidate audit point, the target first reference audit point is used as the reference audit point.
3. The method according to claim 1, characterized in that The candidate review rules include information to be filled in, and the target review rule corresponding to the target bill field is determined based on each of the candidate review rules, including: Displaying each of the candidate audit points, and in response to a selection operation on a target candidate audit point among the candidate audit points, displaying each of the candidate audit rules corresponding to the target candidate audit point; In response to a selection operation on a target alternative review rule among the alternative review rules, the information to be filled in corresponding to the target alternative review rule is displayed, and in response to an editing operation on the information to be filled in, the target review rule corresponding to the target bill field is obtained.
4. The method according to claim 3, characterized in that The audit rule includes rule framework information and rule customization information. The generating of the alternative audit rule corresponding to the alternative audit point based on the audit rule corresponding to the reference audit point includes: Using the rule framework information in the audit rule corresponding to the reference audit point as the rule framework information corresponding to the candidate audit point; According to the rule-defined information in the audit rule corresponding to the reference audit point, determine the information to be filled in and the prompt information corresponding to the alternative audit point, wherein the prompt information is used to prompt the user to fill in the information to be filled in based on the rule-defined information; The alternative review rules are generated according to the rule framework information, the information to be filled in and the prompt information.
5. The method according to claim 4, characterized in that The audit rules correspond to rule types, and the rule types include value range rule types, text matching rule types, and text content rule types; In the case where the rule type is the numerical range rule type, the rule custom information is a numerical range; In the case where the rule type is the text matching rule type, the rule custom information is the text to be matched; In the case where the rule type is the text content rule type, the rule custom information is text semantic information.
6. The method according to claim 2, characterized in that In the case that the target first reference audit point does not exist, the method further includes: A basic audit rule matching the data type information of the candidate audit point is obtained, and an alternative audit rule corresponding to the candidate audit point is generated according to the basic audit rule.
7. A bill auditing device, characterized in that: The device comprises: The first determination module is used to determine the target bill field corresponding to the bill to be reviewed, and when the target bill field does not have a corresponding review rule, determine the reference bill field corresponding to the target bill field in each preset bill field; A second determination module is used to determine each candidate audit point corresponding to the to-be-audited bill based on the bill information contained in the to-be-audited bill, and respectively determine the audit point information corresponding to each of the candidate audit points, wherein the audit point information includes semantic information and data type information; A generating module, for determining, for any of the candidate audit points, a reference audit point from each of the audit points according to the similarity between the audit point information corresponding to the candidate audit point and the audit point information corresponding to each audit point in the reference bill field, and generating an alternative audit rule corresponding to the candidate audit point based on the audit rule corresponding to the reference audit point; The third determination module is used to determine the target audit rule corresponding to the target bill field based on each of the candidate audit rules, and audit the bill to be audited based on the target audit rule.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.
10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.