Waybill weight information anomaly identification method and device, electronic equipment and storage medium

Abstract

The application provides a way, device, electronic equipment and computer readable storage medium for identifying abnormal weight information of a waybill. The way for identifying abnormal weight information of a waybill comprises: obtaining target waybill feature information of a waybill to be identified, which comprises multiple waybill information of the waybill to be identified; obtaining a target feature representation value of the waybill to be identified based on the target waybill feature information; and determining target weight abnormal information of the waybill to be identified according to the target feature representation value and a preset abnormal mapping parameter, wherein the abnormal mapping parameter is used to reflect a constraint relationship between a feature representation value of a waybill and weight abnormal information of the waybill, and the abnormal mapping parameter is obtained by learning sample feature representation values of sample waybills and actual weight abnormal information of the sample waybills. In the application, the identification accuracy of weight abnormal information can be improved, and thus the identification rate of waybills with abnormal weight information can be improved.

Description

Technical Field

[0001] This application relates to the field of data processing technology, specifically to a method, apparatus, electronic device, and computer-readable storage medium for identifying anomalies in waybill weight information. Background Technology

[0002] Weight information is a crucial indicator in the waybill transportation process; for example, the weight of a waybill is an important metric for measuring waybill charges. However, in actual business scenarios, anomalies in waybill weight information, such as under-calculation, often occur.

[0003] Anomalies in waybill weight information can be detected by randomly checking the weight of items on scales during certain operational steps. However, there are various reasons for abnormal waybill weight information, and random checking is time-consuming and has a relatively low detection rate for waybills with abnormal weight information.

[0004] In existing technologies, waybills with abnormal weight information are identified by determining whether the object of the operation on the waybill is a high-risk object, or by determining whether the waybill or the recipient of the shipment is a high-risk recipient. However, in practical applications, the inventors of this application have found that simply identifying waybill weight information abnormalities based on high-risk operation objects or high-risk recipients still results in a relatively low identification rate. Summary of the Invention

[0005] This application provides a method, apparatus, electronic device, and computer-readable storage medium for identifying abnormal weight information on waybills, aiming to solve the problem that the identification rate of waybills with abnormal weight information is relatively low when relying solely on high-risk operation objects or high-risk sender objects to identify abnormal weight information.

[0006] In a first aspect, this application provides a method for identifying anomalies in waybill weight information, the method comprising:

[0007] Obtain the target waybill feature information of the waybill to be identified, wherein the target waybill feature information includes multiple waybill information of the waybill to be identified;

[0008] Based on the target waybill feature information, obtain the target feature representation value of the waybill to be identified;

[0009] Based on the target feature representation value and the preset anomaly mapping parameters, the target weight anomaly information of the waybill to be identified is determined. The anomaly mapping parameters are used to reflect the constraint relationship between the feature representation value of the waybill and the weight anomaly information of the waybill. The anomaly mapping parameters are learned by learning the sample feature representation value and the actual weight anomaly information of the sample waybill.

[0010] Secondly, this application provides a waybill weight information anomaly identification device, the waybill weight information anomaly identification device comprising:

[0011] The first acquisition unit is used to acquire target waybill feature information of the waybill to be identified, wherein the target waybill feature information includes multiple waybill information of the waybill to be identified;

[0012] The second acquisition unit is used to acquire the target feature representation value of the waybill to be identified based on the target waybill feature information;

[0013] The identification unit is used to determine the target weight anomaly information of the waybill to be identified based on the target feature representation value and the preset anomaly mapping parameters. The anomaly mapping parameters are used to reflect the constraint relationship between the feature representation value of the waybill and the weight anomaly information of the waybill. The anomaly mapping parameters are learned by learning the sample feature representation value and the actual weight anomaly information of the sample waybill.

[0014] Thirdly, this application also provides an electronic device, which includes a processor and a memory, wherein the memory stores a computer program, and when the processor calls the computer program in the memory, it executes the steps in any of the waybill weight information anomaly identification methods provided in this application.

[0015] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, the computer program being loaded by a processor to execute the steps in the method for identifying abnormal waybill weight information.

[0016] This application obtains the target feature representation value of the waybill to be identified by combining the target waybill feature information; based on the target feature representation value and preset anomaly mapping parameters, it determines the target weight anomaly information of the waybill to be identified. Firstly, it avoids solely relying on high-risk operation objects or high-risk sender objects to identify waybill weight information anomalies; by combining multiple waybill information from the waybill to be identified, the accuracy of weight anomaly identification can be improved. Secondly, since weight information anomalies can be identified based on multiple waybill information from the waybill to be identified, the waybill identification rate for weight information anomalies can be improved. Thirdly, since the anomaly mapping parameters are learned through sample feature representation values ​​and actual weight anomaly information of sample waybills, the anomaly mapping parameters reflect the constraint relationship between the feature representation values ​​of the waybill and the weight anomaly information of the waybill. Therefore, determining the target weight anomaly information through the anomaly mapping parameters can more accurately identify waybill weight information anomalies. It is evident that the embodiments of this application can improve the accuracy of weight anomaly information identification, thereby improving the waybill identification rate for weight information anomalies. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of a scenario for the waybill weight information anomaly identification system provided in this application embodiment;

[0019] Figure 2 This is a flowchart illustrating a method for identifying anomalies in waybill weight information provided in an embodiment of this application.

[0020] Figure 3 This is a flowchart illustrating one method for learning anomaly mapping parameters in an embodiment of this application;

[0021] Figure 4 This is a schematic diagram illustrating the output of prompt information provided in an embodiment of this application;

[0022] Figure 5 This is a schematic diagram of a target terminal outputting a prompt message according to an embodiment of this application;

[0023] Figure 6 This is another schematic diagram of the target terminal outputting prompt information provided in the embodiments of this application;

[0024] Figure 7 This is a schematic diagram of an embodiment of the waybill weight information anomaly identification device provided in this application.

[0025] Figure 8 This is a schematic diagram of an embodiment of the electronic device provided in this application. Detailed Implementation

[0026] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0027] In the description of the embodiments of this application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, features defined with "first" and "second" may explicitly or implicitly include one or more of the stated features. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0028] To enable any person skilled in the art to implement and use this application, the following description is provided. In this description, details are set forth for purposes of explanation. It should be understood that those skilled in the art will recognize that this application can be implemented without using these specific details. In other instances, well-known processes will not be described in detail to avoid obscuring the description of the embodiments of this application with unnecessary detail. Therefore, this application is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in the embodiments of this application.

[0029] The execution subject of the waybill weight information anomaly identification method in this application embodiment can be the waybill weight information anomaly identification device provided in this application embodiment, or different types of electronic devices such as server equipment, physical host, or user equipment (UE) that integrate the waybill weight information anomaly identification device. The waybill weight information anomaly identification device can be implemented in hardware or software. The UE can be a terminal device such as a smartphone, tablet computer, laptop computer, handheld computer, desktop computer, or personal digital assistant (PDA).

[0030] The electronic device can operate independently or in a cluster.

[0031] See Figure 1 , Figure 1This is a schematic diagram of a scenario for a waybill weight information anomaly identification system provided in this application embodiment. The waybill weight information anomaly identification system may include an electronic device 100, which integrates a waybill weight information anomaly identification device. For example, the electronic device can acquire target waybill feature information of the waybill to be identified, the target waybill feature information including multiple waybill information of the waybill to be identified; based on the target waybill feature information, acquire the target feature representation value of the waybill to be identified; and determine the target weight anomaly information of the waybill to be identified according to the target feature representation value and a preset anomaly mapping parameter, wherein the anomaly mapping parameter is used to reflect the constraint relationship between the feature representation value of the waybill and the weight anomaly information of the waybill, and the anomaly mapping parameter is learned by learning the sample feature representation value and the actual weight anomaly information of the sample waybill.

[0032] In addition, such as Figure 1 As shown, the waybill weight information anomaly identification system may also include a memory 200 for storing data, such as waybill feature information.

[0033] It should be noted that, Figure 1 The schematic diagram of the waybill weight information anomaly identification system shown is merely an example. The waybill weight information anomaly identification system and scenario described in this application embodiment are for the purpose of more clearly illustrating the technical solutions of this application embodiment and do not constitute a limitation on the technical solutions provided by this application embodiment. As those skilled in the art will know, with the evolution of waybill weight information anomaly identification systems and the emergence of new business scenarios, the technical solutions provided by the embodiments of this invention are also applicable to similar technical problems.

[0034] The following describes the method for identifying abnormal waybill weight information provided in the embodiments of this application. In the embodiments of this application, an electronic device is used as the execution subject. For the sake of simplicity and ease of description, the execution subject will be omitted in the subsequent method embodiments.

[0035] Reference Figure 2 , Figure 2 This is a flowchart illustrating a method for identifying anomalies in waybill weight information provided in an embodiment of this application. It should be noted that although the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than that shown here. The method for identifying anomalies in waybill weight information includes steps 201 to 203, wherein:

[0036] 201. Obtain the target waybill feature information of the waybill to be identified.

[0037] The target waybill feature information comprises the waybill information of the waybill to be identified, such as the basic feature information of the waybill to be identified, the combined feature information of the basic feature information, and the attribution feature information of the waybill to be identified. The target waybill feature information includes multiple waybill information of the waybill to be identified.

[0038] Among them, basic feature information refers to the basic information of the waybill to be identified, such as the sender, the pickup, the receiving point, the type of receiving point, the product type, the chargeable weight, and the consigned item.

[0039] Combined feature information refers to information obtained by combining at least two of the basic information of the waybill to be identified, such as pickup object_sender object, consigned item_chargeable weight, and product type_chargeable weight.

[0040] Attribution information refers to the object to which the waybill belongs, such as the sender, pick-up point, or receiving point of the waybill.

[0041] 202. Based on the target waybill feature information, obtain the target feature representation value of the waybill to be identified.

[0042] The target feature representation value is an indication of the waybill information used to predict whether the weight information of the waybill to be identified is abnormal. The target feature representation value can take various forms, including, for example:

[0043] I. The target feature representation value is the preset target anomaly probability corresponding to the target waybill feature information.

[0044] At this point, step A1 is included before step 202:

[0045] A1. Obtain the preset anomaly probability corresponding to the feature information of each waybill.

[0046] At this point, step 202 may specifically include step 2021A:

[0047] 2021A. Obtain the target anomaly probability corresponding to the target waybill feature information from the preset anomaly probabilities corresponding to the feature information of each waybill, and use it as the target feature representation value of the waybill to be identified.

[0048] Since waybill feature information can take many forms in different business scenarios, the following example illustrates how to obtain the preset anomaly probability of waybill feature information in step A1 when the waybill feature information is basic feature information, combined feature information of basic feature information, or attribution feature information.

[0049] (1) Waybill feature information is basic feature information.

[0050] For example, step A1 may specifically include: counting the number of duplicate items and the number of duplicate and abnormal items corresponding to each basic feature information within a preset time period; and determining the abnormal probability corresponding to each basic feature information based on the number of duplicate items and the number of duplicate and abnormal items corresponding to each basic feature information, so as to serve as the preset abnormal probability corresponding to the waybill feature information.

[0051] Taking the basic feature information specifically as the receiving outlet and the waybill feature information corresponding to the preset anomaly probability as the anomaly probability of each receiving outlet as an example, where the anomaly probability of the receiving outlet is the ratio of the number of weight-abnormal items to the number of duplicate items at the receiving outlet, step A1 can specifically include: statistically analyzing the number of duplicate items and the number of duplicate-abnormal items at each receiving outlet within a preset time period; and determining the anomaly probability of each receiving outlet based on the number of duplicate items and the number of duplicate-abnormal items at each receiving outlet, thus obtaining the anomaly probability of each receiving outlet.

[0052] The quantity of duplicated items refers to the number of waybills whose weights have been verified.

[0053] The number of waybills with weight discrepancies refers to the number of waybills that have undergone weight verification but contain weight discrepancies such as under-calculation of chargeable weight.

[0054] Here, the probability of an anomaly is the ratio of the number of duplicated abnormal parts to the number of duplicated parts.

[0055] For example, by statistically analyzing the number of duplicate and abnormal items at each receiving point within a preset time period, and based on the number of duplicate and abnormal items at each receiving point, the abnormal probability of each receiving point can be determined, resulting in the preset abnormal probabilities corresponding to each waybill feature information as shown in Table 1 below.

[0056] Table 1

[0057] Waybill Feature Information (Receiving Location) abnormal probability Branch 1 Probability 1 Branch 2 Probability 2 … … outlet n probability n

[0058] At this point, the target waybill feature information can be the target receiving point of the waybill to be identified. Step 2021A can specifically include: obtaining the target abnormal probability of the target receiving point from the abnormal probabilities of each receiving point, and using it as the target feature representation value of the waybill to be identified.

[0059] Similarly, just as the basic feature information is used to determine the probability of anomalies when it is the receiving point, the basic feature information can be the sender object, the pickup object, the receiving point type, the product type, the chargeable weight, and the consigned item. The probability of anomalies for each sender object, each pickup object, each receiving point type, each product type, the chargeable weight, and the consigned item can be determined respectively. The resulting probability of anomalies for each basic feature information can be used as the preset probability of anomalies for the waybill feature information.

[0060] Similarly, the target waybill feature information can also be the target sender, target pickup, target receiving point type, target product type, target chargeable weight, or target consigned item of the waybill to be identified. Step 2021A can also be implemented by replacing "receiving point" in the above-mentioned "obtaining the target anomaly probability of the target receiving point from the anomaly probabilities of each receiving point, and using it as the target feature representation value of the waybill to be identified" with "sender," "pickup," "receiving point type," "product type," "chargeable weight," or "consigned item," etc., to determine the target feature representation value of the waybill to be identified.

[0061] Furthermore, to improve the comprehensiveness of feature information and thus enhance the accuracy of anomaly identification in waybill weight information, the anomaly probabilities of at least two of the basic feature information, such as sender, pick-up, receiving point, receiving point type, product type, chargeable weight, and consigned item, can be combined to determine the target feature representation value of the waybill to be identified. In this case, step 2021A may specifically include: obtaining the anomaly probability of each target basic feature information of the waybill to be identified from the preset anomaly probabilities corresponding to each basic feature information; fusing and representing the anomaly probabilities of each target basic feature information to obtain the target anomaly probability corresponding to the basic feature information of the waybill to be identified, which serves as the target feature representation value of the waybill to be identified.

[0062] For example, the target feature representation value can be determined by simultaneously combining seven basic feature information of the waybill to be identified: the target sender, the target pickup, the target receiving point, the target receiving point type, the target product type, the target chargeable weight, and the target consigned item. First, using the above method, the anomalous probabilities h1, h2, h3, h4, h5, h6, and h7 of the target sender, pickup, receiving point, and receiving point type can be determined respectively. Then, the anomalous probabilities h1, h2, h3, h4, h5, h6, and h7 are merged and represented into an array {h1, h2, h3, h4, h5, h6, h7}, yielding the target anomalous probabilities {h1, h2, h3, h4, h5, h6, h7} corresponding to the basic feature information of the waybill to be identified, which serve as the target feature representation value of the waybill to be identified.

[0063] By combining multiple basic feature information of multiple waybills to be identified to determine the target feature representation value, the target feature representation value can integrate information from multiple aspects, providing comprehensive data for the subsequent identification of abnormal waybill weight information, thereby improving the accuracy of the identification of abnormal waybill weight information.

[0064] (2) Waybill feature information is a combination of basic feature information.

[0065] For example, step A1 may specifically include: counting the number of duplicate items and the number of duplicate and abnormal items corresponding to each combination of feature information within a preset time period; and determining the abnormal probability corresponding to each combination of feature information based on the number of duplicate items and the number of duplicate and abnormal items, so as to use the preset abnormal probability corresponding to the waybill feature information.

[0066] Taking the combined feature information specifically as the pickup object_sending object and the waybill feature information corresponding to the preset abnormal probability as the abnormal probability of each pickup object_sending object as an example. Wherein, the abnormal probability of a pickup object_sending object is the ratio of the number of items with abnormal weight to the number of items with duplicate weight corresponding to the pickup object_sending object. In this case, step A1 may specifically include: statistically analyzing the number of duplicate weight items and the number of duplicate weight abnormal items for each pickup object_sending object within a preset time period; based on the number of duplicate weight items and the number of duplicate weight abnormal items for each pickup object_sending object, determining the abnormal probability of each pickup object_sending object, thus obtaining the abnormal probability of each pickup object_sending object.

[0067] The quantity of duplicated items refers to the number of waybills whose weights have been verified.

[0068] The number of waybills with weight discrepancies refers to the number of waybills that have undergone weight verification but contain weight discrepancies such as under-calculation of chargeable weight.

[0069] Here, the probability of an anomaly is the ratio of the number of duplicated abnormal parts to the number of duplicated parts.

[0070] For example, by statistically analyzing the number of duplicate items and the number of duplicate / abnormal items for each pickup / sending object within a preset time period, and based on the number of duplicate / abnormal items for each pickup / sending object, the abnormal probability of each pickup / sending object can be determined, resulting in the preset abnormal probabilities corresponding to each waybill feature information as shown in Table 2 below.

[0071] Table 2

[0072]

[0073] At this point, the target waybill feature information can be the target pickup object_target sender object of the waybill to be identified. Step 2021A can specifically include: obtaining the target anomaly probability of the target pickup object_target sender object from the anomaly probabilities of each pickup object_sender object, and using it as the target feature representation value of the waybill to be identified.

[0074] Similarly, similar to the method of determining the anomaly probability when the combined feature information is the parcel_sending object, we can determine the anomaly probability of each parcel_billed weight and each product_billed weight by taking the combined feature information as the parcel_billed weight and the product_billed weight respectively, and obtain the anomaly probability corresponding to each combined feature information, which can be used as the preset anomaly probability corresponding to the waybill feature information.

[0075] Similarly, the target waybill feature information can also be the target consignment_target chargeable weight, or the target product type_target chargeable weight, etc., of the waybill to be identified. Step 2021A can also be implemented by replacing "consignment_sending object" with "consignment_chargeable weight" or "product type_chargeable weight," etc., in the above-mentioned "obtain the target anomaly probability of the target consignment_target sending object from the anomaly probabilities of each consignment_sending object, and use it as the target feature representation value of the waybill to be identified," to determine the target feature representation value of the waybill to be identified.

[0076] Furthermore, to improve the comprehensiveness of feature information and thus enhance the accuracy of anomaly identification in waybill weight information, the anomaly probabilities of at least two of the combined feature information, such as pickup object_sender object, consigned item_chargeable weight, and product type_chargeable weight, can be combined to determine the target feature representation value of the waybill to be identified. In this case, step 2021A may specifically include: obtaining the anomaly probability of each target combined feature information of the waybill to be identified from the preset anomaly probabilities corresponding to each combined feature information; fusing and representing the anomaly probabilities of each target combined feature information to obtain the target anomaly probability corresponding to the combined feature information of the waybill to be identified, which is then used as the target feature representation value of the waybill to be identified.

[0077] For example, the target feature representation value can be determined by simultaneously combining three combined feature information of the waybill to be identified: pickup object_sender object, consigned item_chargeable weight, and product type_chargeable weight. First, using the above method, the anomalous probability h1 of the target pickup object_target sender object, the anomalous probability h2 of the target consigned item_target chargeable weight, and the anomalous probability h3 of the target product type_target chargeable weight can be determined separately. Then, the anomalous probabilities h1, h2, and h3 are merged and represented into an array {h1, h2, h3}, resulting in the target anomalous probability {h1, h2, h3} corresponding to the combined feature information of the waybill to be identified, which serves as the target feature representation value of the waybill to be identified.

[0078] By combining multiple feature information from various waybills to be identified, the target feature representation value is determined, allowing for the fusion of multi-faceted information. Since the fused feature information can start from the low dimension of basic feature information and construct higher-dimensional feature information to characterize high-risk features, such as combining the consigned item and chargeable weight of a waybill, it more closely reflects the risk characteristics of weight information anomalies in actual business scenarios. This provides comprehensive data for subsequent identification of waybill weight information anomalies, enabling more accurate prediction of waybill weight information anomalies based on actual business conditions, thereby improving the accuracy of waybill weight information anomaly identification.

[0079] (3) Waybill feature information is attribution feature information.

[0080] For example, step A1 may specifically include: counting the number of duplicate shipments of each attributed feature information and the number of duplicate and abnormal shipments of each attributed feature information corresponding to the receiving outlet within a preset time period; and determining the abnormal probability of each attributed feature information based on the number of duplicate shipments and the number of duplicate and abnormal shipments of each attributed feature information, so as to use as the preset abnormal probability corresponding to the waybill feature information.

[0081] Taking the example where the attribution feature information specifically refers to the target receiving point and the waybill feature information corresponds to the preset anomaly probability of the target receiving point, the anomaly probability of the target receiving point is defined as the ratio of the number of weight-abnormal items to the number of duplicate items in the target waybill for that receiving point. Here, the target receiving point refers to the receiving point corresponding to a waybill that meets preset conditions (e.g., less than a preset weight threshold and no package creation operation); the target waybill is a waybill among the receiving points of the target receiving point that meets the preset conditions (e.g., less than a preset weight threshold and no package creation operation, e.g., <3kg and no package creation operation). At this point, step A1 may specifically include: counting the number of duplicate and heavy items in the target waybill of each target receiving point within a preset time period, and the number of duplicate and heavy abnormal items in the target waybill of each target receiving point; based on the number of duplicate and heavy items in the target waybill of each target receiving point, and the number of duplicate and heavy abnormal items in the target waybill of each target receiving point, determining the abnormal probability of the target waybill of each target receiving point, and obtaining the abnormal probability of each target receiving point.

[0082] The preset conditions here are for illustrative purposes only. The specific conditions can be adjusted according to the actual business scenario requirements and are not limited to these conditions.

[0083] To facilitate logistics management, there is usually a package creation operation in the logistics field to prevent the loss of waybills that are small in weight or volume. The package creation operation is to pack multiple waybills that are small in weight or volume into a package, and then transport them as a single package.

[0084] The quantity of duplicated items refers to the number of waybills whose weights have been verified.

[0085] The number of waybills with weight discrepancies refers to the number of waybills that have undergone weight verification but contain weight discrepancies such as under-calculation of chargeable weight.

[0086] Here, the probability of an anomaly is the ratio of the number of duplicated abnormal parts to the number of duplicated parts.

[0087] For example, by statistically analyzing the number of duplicate and repetitive items and the number of duplicate and repetitive abnormal items of the target waybills at each target receiving point within a preset time period, and based on the number of duplicate and repetitive items and the number of duplicate and repetitive abnormal items of the target waybills at each target receiving point, the abnormal probability of the target waybill at each target receiving point can be determined, as shown in Table 3 below.

[0088] Table 3

[0089] Target receiving point abnormal probability Branch 1 Probability 1 Branch 2 Probability 2 … … outlet n probability n

[0090] At this time, the target waybill feature information can be the target receiving point of the waybill to be identified. Step 2021A can specifically include: obtaining the target anomaly probability of the target waybill of the target receiving point of the waybill to be identified from the anomaly probability of the target waybill of each target receiving point, and using it as the target feature representation value of the waybill to be identified.

[0091] By combining multiple attribution feature information from various waybills to be identified, the target feature representation value is determined, allowing for the integration of multi-faceted information. Since high-risk features relevant to actual business scenarios can be constructed based on attribution feature information, such as the probability of anomalies in target waybills whose corresponding network points meet preset conditions (e.g., weight less than a preset threshold and no package creation operation), this more closely reflects the risk characteristics of weight information anomalies in actual business scenarios. This provides comprehensive data for subsequent identification of waybill weight information anomalies, enabling more accurate prediction of waybill weight information anomalies based on actual business conditions, thereby improving the accuracy of waybill weight information anomaly identification.

[0092] II. The target feature representation value is the target derived representation value corresponding to the target waybill feature information. The derived representation value is determined based on the target anomaly probability.

[0093] There are several ways to determine the target feature representation value based on the target derived representation value. For example, the target derived representation value can be stored in a preset database, and in step 202, the target derived representation value can be directly obtained as the target feature representation value. Alternatively, the target anomaly probability can be stored in a preset database, and in step 202, the pre-stored target anomaly probability can be obtained to determine the target derived representation value in real time, and then the real-time determined target derived representation value can be used as the target feature representation value. Examples (1) and (2) are given below for illustration:

[0094] (1) In some embodiments, the target derived representation value corresponding to the target waybill feature information is stored in a preset database, and in step 202, the target derived representation value corresponding to the target waybill feature information can be directly obtained as the target feature representation value. In this case, steps B1 to B2 are included before step 202:

[0095] B1. Obtain the preset anomaly probability corresponding to the feature information of each waybill.

[0096] Step B1 is similar to step A1 above, and you can refer to the description of step A1 above for details, which will not be repeated here.

[0097] B2. Based on the preset anomaly probability corresponding to each waybill feature information, determine the preset derived representation value corresponding to each waybill feature information.

[0098] The derived representation value is an extended representation of the waybill feature information.

[0099] For example, as shown in Table 4, firstly, based on the preset anomaly probability corresponding to each waybill feature information, the lower quartile of the preset anomaly probability corresponding to each waybill feature information is determined; then, the first preset value is used as the derived representation value of waybill feature information whose (corresponding preset anomaly probability) is less than the lower quartile, and the second preset value is used as the derived representation value of waybill feature information whose (corresponding preset anomaly probability) is greater than or equal to the lower quartile, thereby obtaining the preset derived representation value corresponding to each waybill feature information.

[0100] Table 4

[0101] Anomaly probability x Derived representation value x < lower quartile First preset value x ≥ lower quartile Second preset value

[0102] For example, taking the waybill feature information as specifically the receiving outlet, the preset anomaly probability corresponding to the waybill feature information as the anomaly probability of each receiving outlet, the first preset value as 1, and the second preset value as 0 as an example. First, based on the preset anomaly probability corresponding to each receiving outlet (e.g., the preset anomaly probabilities corresponding to outlets 1, 2, and 3 are 9%, 20%, 30%, and 40%, respectively), the lower quartile (e.g., 10%) of the preset anomaly probability corresponding to each receiving outlet is determined; then, the first preset value is used as the derived representation value of receiving outlets less than the lower quartile, and the second preset value is used as the derived representation value of receiving outlets greater than or equal to the lower quartile, thereby obtaining the preset derived representation value corresponding to each receiving outlet (i.e., the preset derived representation values ​​corresponding to outlets 1, 2, and 3 are 1, 0, and 0, respectively).

[0103] The first and second preset values ​​here are just examples. The specific values ​​of the first and second preset values ​​can be determined according to actual business needs and are not limited to this.

[0104] Here, the lower quartile is just an example of "determining the preset derived representation value corresponding to the feature information of each waybill". In practice, the lower quartile can be replaced with the lower n quartile, or with the upper quartile, or with the upper n quartile according to actual business needs. It is not limited to this example.

[0105] Correspondingly, step 202 may specifically include steps 2021B to 2022B:

[0106] 2021B. Obtain the target derived representation value corresponding to the target waybill feature information from the preset derived representation values ​​corresponding to each waybill feature information, and use it as the target feature representation value.

[0107] 2022B. Based on the target derived representation value, determine the target feature representation value.

[0108] (2) In other embodiments, the target waybill feature information corresponds to a preset target anomaly probability stored in a preset database. A preset probability relationship exists between the target anomaly probability and the target derived representation value. In step 202, the target derived representation value is determined in real time based on the target anomaly probability, and then the target feature representation value is determined based on the target derived representation value. In this case, step C1 is included before step 202:

[0109] C1. Obtain the preset anomaly probability corresponding to the feature information of each waybill.

[0110] Step C1 is similar to step A1 above, and the details can be found in the description of step A1 above, which will not be repeated here.

[0111] Correspondingly, step 202 may specifically include steps 2021C to 2023C:

[0112] 2021C. Obtain the target anomaly probability corresponding to the target waybill feature information from the preset anomaly probabilities corresponding to each waybill feature information.

[0113] In particular, step 2021C, which obtains the target anomaly probability corresponding to the target waybill feature information, is similar to obtaining the target anomaly probability corresponding to the target waybill feature information in step 2021A above. For details, please refer to the description of step 2021A above, which will not be repeated here.

[0114] 2022C. Based on the relationship between the target anomaly probability and the preset representation value probability, determine the target derived representation value of the waybill to be identified.

[0115] Among them, the preset representation value probability relationship is used to indicate the relationship between the target anomaly probability and the target derived representation value.

[0116] For example, the preset representation value probability relationship is shown in Table 5 below. If the target anomaly probability is determined to be x1, then the target derived representation value of the waybill to be identified can be determined to be y1.

[0117] Table 5

[0118] Anomaly probability x Derived representation value y x1 y1 x2 y2 … … xn yn

[0119] 2023C. The target derived representation value is used as the target feature representation value.

[0120] At this point, the target feature representation value can be directly used as the target feature representation value. For ease of understanding, let's continue with the example from step 2022C above. For instance, when the target derived representation value of the waybill to be identified is determined to be y1, the target derived representation value y1 can be directly used as the target feature representation value.

[0121] Furthermore, in order to improve the comprehensiveness of the target feature representation value and thus improve the accuracy of anomaly identification of waybill weight information, referring to the above steps 2021B or steps 2021C to 2023C for obtaining derived feature values, multiple target derived representation values ​​can be determined based on multiple basic feature information in the target waybill feature information, and used as target feature representation values.

[0122] For example, basic feature information includes the target sender, target pickup, target receiving point, target receiving point type, target product type, target chargeable weight, and target consigned item for the waybill to be identified. The target derived representation values ​​q1, q2, q3, q4, q5, q6, and q7 of the target sender, pickup, receiving point, receiving point type, product type, chargeable weight, and consigned item can be determined separately. Then, the multiple target derived representation values ​​q1, q2, q3, q4, q5, q6, and q7 determined based on the basic feature information are merged and represented into an array {q1,q2,q3,q4,q5,q6,q7}, which serves as the target feature representation value for the waybill to be identified.

[0123] Alternatively, multiple target-derived representation values ​​can be determined based on multiple combined feature information in the target waybill feature information, and used as target feature representation values.

[0124] For example, basic feature information includes the target pickup object_target sender object, target consigned item_target chargeable weight, and target product type_target chargeable weight of the waybill to be identified. The target derived representation values ​​q1, q2, and q3 of the target pickup object_target sender object, target consigned item_target chargeable weight, and target product type_target chargeable weight can be determined separately. Then, the multiple target derived representation values ​​q1, q2, and q3 determined based on the fused feature information are fused and represented into an array {q1,q2,q3}, which serves as the target feature representation value of the waybill to be identified.

[0125] Alternatively, multiple target-derived representation values ​​can be determined based on multiple attribution feature information in the target waybill feature information, and used as target feature representation values.

[0126] Alternatively, based on at least two of the basic feature information, combined feature information, and subordinate feature information in the target waybill feature information, at least two target derived representation values ​​can be determined as target feature representation values.

[0127] III. The target feature representation values ​​are the preset target anomaly probability and target derived representation values ​​corresponding to the target waybill feature information.

[0128] Furthermore, in order to improve the comprehensiveness of the target feature representation value and thus improve the accuracy of anomaly identification of waybill weight information, the target anomaly probability and the target derived representation value can be combined simultaneously as the target feature representation value.

[0129] There are several ways to determine the target feature representation value by combining the target anomaly probability and the target derived representation value. For example, similar to the above "II.", the target anomaly probability and the target derived representation value can be stored in a preset database at the same time, and can be directly obtained in step 202 to determine the target feature representation value. Alternatively, the target anomaly probability can be stored in a preset database, and the target derived representation value can be determined in real time by obtaining the pre-stored target anomaly probability in step 202, and then the target feature representation value can be determined by combining the target anomaly probability and the target derived representation value. Examples (1) and (2) are given below for illustration:

[0130] (1) In some embodiments, the target anomaly probability and target derived representation value corresponding to the target waybill feature information are stored in a preset database. In step 202, the target derived representation value corresponding to the target waybill feature information can be directly obtained as the target feature representation value. In this case, steps D1 to D1, obtaining the preset anomaly probability corresponding to each waybill feature information, are also included before step 202.

[0131] D2. Based on the preset anomaly probability corresponding to each waybill feature information, determine the preset derived representation value corresponding to each waybill feature information.

[0132] Steps D1 to D2 are similar to steps B1 to B2 above. For details, please refer to the description of steps B1 to B2 above. They will not be repeated here.

[0133] Correspondingly, step 202 may specifically include steps 2021D to 2023D:

[0134] 2021D. Obtain the target anomaly probability corresponding to the target waybill feature information from the preset anomaly probabilities corresponding to each waybill feature information.

[0135] In particular, step 2021D, which obtains the target anomaly probability corresponding to the target waybill feature information, is similar to obtaining the target anomaly probability corresponding to the target waybill feature information in step 2021A above. For details, please refer to the description of step 2021A above, which will not be repeated here.

[0136] 2022D. Obtain the target derived representation value corresponding to the target waybill feature information from the preset derived representation values ​​corresponding to each waybill feature information.

[0137] 2023D. Based on the target anomaly probability and the target derived representation value, determine the target feature representation value.

[0138] Specifically, the target anomaly probability and the target derived representation value can be fused and represented as the target feature representation value.

[0139] For example, the target waybill feature information consists of seven basic features: the target sender, the target pickup, the target receiving point, the target receiving point type, the target product type, the target chargeable weight, and the target consigned item.

[0140] Based on these seven basic characteristics, the following can be determined sequentially: the abnormal probability h1 of the target sender, the abnormal probability h2 of the target pickup, the abnormal probability h3 of the target receiving point, the abnormal probability h4 of the target receiving point type, the abnormal probability h5 of the target product type, the abnormal probability h6 of the target chargeable weight, the abnormal probability h7 of the target consigned item, and the target derived representation values ​​q1, q2, q3, q4, q5, q6, and q7 of the target sender, pickup, receiving point, receiving point type, product type, chargeable weight, and consigned item. At this point, the anomaly probabilities h1, h2, h3, h4, h5, h6, h7 and the target derived representation values ​​q1, q2, q3, q4, q5, q6, q7 can be merged into an array {h1, h2, h3, h4, h5, h6, h7, q1, q2, q3, q4, q5, q6, q7}, which can be used as the target feature representation value of the waybill to be identified.

[0141] (2) In some embodiments, the target anomaly probability is stored in a preset database, and there is a preset probability relationship between the target anomaly probability and the target derived representation value. In step 202, the target derived representation value is determined in real time based on the target anomaly probability, and then the target feature representation value is determined by combining the target anomaly probability and the real-time determined target derived representation value. In this case, step E1 is included before step 202:

[0142] E1. Obtain the preset anomaly probability corresponding to the feature information of each waybill.

[0143] Step E1 is similar to step A1 above, and you can refer to the description of step A1 above for details, which will not be repeated here.

[0144] Correspondingly, step 202 may specifically include steps 2021E to 2023E:

[0145] 2021E: Obtain the target anomaly probability corresponding to the target waybill feature information from the preset anomaly probabilities corresponding to each waybill feature information.

[0146] In particular, step 2021E, which obtains the target anomaly probability corresponding to the target waybill feature information, is similar to obtaining the target anomaly probability corresponding to the target waybill feature information in step 2021A above. For details, please refer to the description of step 2021A above, which will not be repeated here.

[0147] 2022E. Based on the relationship between the target anomaly probability and the preset representation value probability, determine the target derived representation value of the waybill to be identified.

[0148] The preset representation value probability relationship is used to indicate the relationship between the target anomaly probability and the target derived representation value.

[0149] Step 2022E is similar to step 2022C above. For details, please refer to the description of step 2022C above. It will not be repeated here.

[0150] 2023E. Based on the target anomaly probability and the target derived representation value, determine the target feature representation value.

[0151] Step 2023E is similar to step 2023D above. For details, please refer to the description of step 2023D above. It will not be repeated here.

[0152] As can be seen from the above steps 2021A, 2021C to 2023C, 2021D to 2023D, and 2021E to 2023E, step 202 can specifically include: obtaining the target anomaly probability corresponding to the target waybill feature information from the preset anomaly probabilities corresponding to each waybill feature information; and determining the target feature representation value based on the target anomaly probability.

[0153] As can be seen from the above, the target waybill feature information is taken from at least one of the basic feature information of the waybill to be identified, the combined feature information of the basic feature information, and the attribution feature information of the waybill to be identified. The target feature representation value of the waybill to be identified can be determined based on at least one of the basic feature information of the waybill to be identified, the combined feature information of the basic feature information, and the attribution feature information of the waybill to be identified. Specifically, refer to the descriptions of steps 2021A, 2021B~2022B, 2021C~2023C, 2021D~2023D, and 2021E~2023E above.

[0154] The following example uses the target feature representation value as derived representation value and anomaly probability, and the target waybill feature information as the basic feature information, combined feature information, and attribution feature information of the waybill to be identified, to illustrate the determination of the target feature representation value. In this case, step 202 may specifically include steps 2021F to 2023F:

[0155] 2021F, obtain the first abnormal probability corresponding to the basic feature information, the second abnormal probability corresponding to the combined feature information, and the third abnormal probability corresponding to the attribution feature information.

[0156] Among them, the first anomaly probability refers to the target anomaly probability corresponding to the target waybill feature information, which is determined based on the basic feature information.

[0157] The second anomaly probability refers to the target anomaly probability corresponding to the target waybill feature information, determined based on the combined feature information.

[0158] The third anomaly probability refers to the target anomaly probability corresponding to the target waybill feature information, determined based on the attribution feature information.

[0159] The determination of the first, second, and third anomaly probabilities is similar to that of the target anomaly probabilities mentioned above. For details, please refer to the relevant explanations in steps 2021A, 2021B to 2022B, 2021C to 2023C, 2021D to 2023D, or 2021E to 2023E. These will not be repeated here.

[0160] 2022F. Based on the first abnormal probability, second abnormal probability, third abnormal probability and preset representation value probability mapping relationship shown, determine the first derived representation value, second derived representation value and third derived representation value of the waybill to be identified.

[0161] The first derived representation value refers to the preset derived representation value corresponding to the target waybill feature information, which is determined based on the first anomaly probability.

[0162] The second derived representation value refers to the preset derived representation value corresponding to the target waybill feature information, which is determined based on the second anomaly probability.

[0163] The third derived representation value refers to the preset derived representation value corresponding to the target waybill feature information, which is determined based on the third anomaly probability.

[0164] The determination of the first derived representation value, the second derived representation value, and the third derived representation value is similar to that of the target derived representation value. For details, please refer to the relevant explanations of steps 2021A, 2021B to 2022B, 2021C to 2023C, 2021D to 2023D, or 2021E to 2023E. These will not be repeated here.

[0165] 2023F. Based on the first anomaly probability, the second anomaly probability, the third anomaly probability, the first derived representation value, the second derived representation value, and the third derived representation value, the target feature representation value of the waybill to be identified is determined.

[0166] Specifically, the first anomaly probability, the second anomaly probability, the third anomaly probability, the first derived representation value, the second derived representation value, and the third derived representation value can be fused and represented as the target feature representation value.

[0167] For example, the first anomaly probability, the second anomaly probability, the third anomaly probability, the first derived representation value, the second derived representation value, and the third derived representation value are h1, h2, h3, q1, q2, and q3, respectively. h1, h2, h3, q1, q2, and q3 can be fused and represented into an array {h1, h2, h3, h4, h5, h6, h7}, which serves as the target feature representation value of the waybill to be identified.

[0168] By combining the basic feature information, fused feature information, and attribution feature information of the waybill to be identified to determine the target feature representation value, the target feature representation value can integrate multiple aspects of information, providing comprehensive data for subsequent identification of waybill weight information anomalies, thereby improving the accuracy of waybill weight information anomaly identification. Furthermore, on the one hand, because fused feature information can start from the low dimension of basic feature information to construct higher-dimensional feature information to characterize high-risk features, such as combining the consigned item and chargeable weight of the waybill, it more closely reflects the risk characteristics of weight information anomalies in actual business scenarios; thus, it can more closely predict the weight information anomalies of waybills in actual business situations, thereby improving the accuracy of waybill weight information anomaly identification. On the other hand, because attribution feature information can construct high-risk features related to actual business scenarios, such as the anomaly probability of target waybills corresponding to network points that meet preset conditions such as weight less than a preset weight threshold and no package creation operation, it more closely reflects the risk characteristics of weight information anomalies in actual business scenarios, thereby improving the accuracy of waybill weight information anomaly identification.

[0169] 203. Determine the target weight anomaly information of the waybill to be identified based on the target feature representation value and the preset anomaly mapping parameters.

[0170] The anomaly mapping parameter is used to reflect the constraint relationship between the feature representation value of the waybill and the weight anomaly information of the waybill. The anomaly mapping parameter is learned by the sample feature representation value of the sample waybill and the actual weight anomaly information of the sample waybill.

[0171] Among them, the target weight anomaly information is used to indicate whether the weight information of the waybill to be identified is abnormal.

[0172] Abnormalities in waybill weight information can include issues such as under-counting. Under-counting refers to a situation where the chargeable weight entered when sending a shipment is lower than the actual chargeable weight and the error is significant. Under-counting on waybills may be due to employee operational errors, tool malfunctions, or human error. Regardless of the cause, it is necessary to identify whether the waybill weight information is abnormal in order to promptly detect abnormalities and facilitate effective waybill management.

[0173] In some embodiments, anomaly mapping parameters can be used to predict the anomaly probability value of the waybill to be identified based on the sample feature representation value, which serves as the predicted weight anomaly information. Specifically, when the anomaly probability value is greater than a preset anomaly threshold, the waybill weight information of the waybill to be identified is determined to be abnormal; when the anomaly probability value is less than or equal to the preset anomaly threshold, the waybill weight information of the waybill to be identified is determined to be normal.

[0174] As can be seen from the above, this application embodiment obtains the target feature representation value of the waybill to be identified by combining the target waybill feature information of the waybill to be identified; based on the target feature representation value and the preset anomaly mapping parameters, the target weight anomaly information of the waybill to be identified is determined. Firstly, it avoids simply relying on high-risk operation objects or high-risk sender objects to identify waybill weight information anomalies; by combining multiple waybill information from the waybill to be identified to identify weight information anomalies, the accuracy of weight anomaly identification can be improved. Secondly, since weight information anomalies can be identified based on multiple waybill information from the waybill to be identified, the waybill identification rate for weight information anomalies can be improved. Thirdly, since the anomaly mapping parameters are learned through the sample feature representation values ​​and actual weight anomaly information of sample waybills, the anomaly mapping parameters reflect the constraint relationship between the feature representation values ​​of the waybill and the weight anomaly information of the waybill; therefore, determining the target weight anomaly information through the anomaly mapping parameters can more accurately identify waybill weight information anomalies. Therefore, this application embodiment can improve the accuracy of weight anomaly information identification, and thus improve the waybill identification rate for weight information anomalies.

[0175] In some embodiments of this application, a method for obtaining the exception mapping parameters is also provided, specifically, as follows: Figure 3 As shown, Figure 3 This is a flowchart illustrating a method for learning anomaly mapping parameters in an embodiment of this application. The preset anomaly mapping parameters can be obtained through the following steps 301 to 305:

[0176] 301. Obtain the training dataset.

[0177] The training dataset includes multiple sample waybills, each labeled with information about anomalies in actual weight.

[0178] Actual weight anomaly information refers to information indicating whether the actual weight information of the sample waybill is abnormal.

[0179] 302. Obtain the sample feature representation value of the sample waybill.

[0180] Among them, the sample feature representation value is an indication of whether the weight information of the sample waybill is abnormal and the waybill information of the sample waybill.

[0181] Specifically, firstly, the sample waybill feature information is obtained; then, based on the sample waybill feature information, the sample feature representation value of the sample waybill is obtained.

[0182] Similar to the target waybill feature information, the sample waybill feature information can also be the waybill information of the sample waybill, such as the basic feature information of the sample waybill, the combined feature information of the basic feature information, and the attribution feature information of the sample waybill.

[0183] The method of obtaining the sample feature representation value based on the sample waybill feature information in step 302 is similar to the method of obtaining the target feature representation value in step 202 above. For details, please refer to the description of step 202 above, which will not be repeated here.

[0184] 303. Using the decision tree to be trained, prediction is made based on the sample feature representation values ​​to obtain the predicted weight anomaly information of the sample waybill.

[0185] Among them, the predicted weight anomaly information refers to the information obtained through prediction that indicates whether the weight information of the sample waybill is abnormal.

[0186] For example, the decision tree to be trained is a GBDT (Gradient Boosting Decision Tree) model. Since the GBDT model can perfectly fit continuous values, it can predict the weight anomaly information of waybills without special processing of continuous values ​​(such as anomaly probabilities) in the waybill scenario.

[0187] The decision tree to be trained here is just an example. It can be adjusted according to actual needs, such as changing the decision tree to be trained to other open source tree models or tree models that may appear in the future. This is not a limitation.

[0188] In some embodiments, the abnormal value of the sample waybill is obtained by predicting the sample waybill based on the sample feature representation value using a decision tree to be trained, and this abnormal value is used as the predicted weight anomaly information. Specifically, when the abnormal value of the waybill is greater than a preset anomaly threshold, the waybill weight information of the sample waybill is determined to be abnormal; when the abnormal value of the waybill is less than or equal to the preset anomaly threshold, the waybill weight information of the sample waybill is determined to be normal.

[0189] 304. Based on the predicted weight anomaly information and the actual weight anomaly information, adjust the model parameters of the decision tree to be trained until the preset stop training condition is met, and obtain the trained decision tree.

[0190] Specifically, step 304 may include: determining the prediction loss of the decision tree to be trained based on the predicted weight anomaly information and the actual weight anomaly information; adjusting the model parameters of the decision tree to be trained based on the prediction loss of the decision tree to be trained until the preset stopping training condition is met, and thus obtaining the trained decision tree.

[0191] For example, a prediction loss function is set for the decision tree to be trained, so that the decision tree can learn the constraint relationship between the feature representation value of the waybill and the weight anomaly information of the waybill. The prediction loss function is set according to the predicted weight anomaly information output by the decision tree to be trained. During the training process, the value of the prediction loss function is the prediction loss of the decision tree to be trained. The prediction loss of the decision tree to be trained can be calculated by substituting the predicted weight anomaly information and the actual weight anomaly information into the prediction loss function. In this embodiment, the specific function type of the prediction loss function is not limited. For example, the prediction loss function can be a logarithmic loss function or an exponential loss function.

[0192] The preset training stop conditions can be set according to actual needs. For example, it could be when the predicted loss is less than a preset value, or when the predicted loss basically stops changing, i.e., the difference in predicted loss between adjacent training iterations is less than the preset value; or when the number of iterations of the decision tree to be trained reaches the maximum number of iterations.

[0193] 305. Extract the model parameters of the trained decision tree to serve as the anomaly mapping parameters.

[0194] Since the trained decision tree is learned from the sample feature representations and actual weight anomaly information of the sample waybills, its model parameters have learned the constraint relationship between the feature representations and the weight anomaly information of the waybills. By extracting the model parameters of the trained decision tree as anomaly mapping parameters to predict the target weight anomaly information of the waybills to be identified, accurate identification of anomalies in the weight information of waybills can be guaranteed.

[0195] Furthermore, to facilitate the management of waybills with abnormal weight information, such as Figure 4 As shown, Figure 4 This is a schematic diagram illustrating the output of the prompt information provided in this application embodiment. After determining the target weight anomaly information of the waybill to be identified based on the target feature representation value and preset anomaly mapping parameters, the method further includes: when it is determined that the weight information of the waybill to be identified is abnormal, outputting a weight anomaly prompt information through the target terminal of the waybill to be identified. Alternatively, when it is determined that the weight information of the waybill to be identified is abnormal, outputting a prompt information suggesting a review of the weight of the waybill to be identified through the target terminal of the waybill to be identified.

[0196] The weight anomaly alert is used to indicate that the weight information of the waybill to be identified is abnormal.

[0197] By having the target terminal of the waybill to be identified, such as a scanning terminal or management platform, output weight abnormality prompt information, relevant staff can promptly detect abnormalities in the weight information of the waybill and thus handle the waybill with abnormal weight information in a timely manner.

[0198] Furthermore, to facilitate the management of waybills with abnormal weight information, such as Figure 4 As shown, after determining the target weight anomaly information of the waybill to be identified based on the target feature representation value and preset anomaly mapping parameters, the method further includes: when it is determined that the weight information of the waybill to be identified is abnormal, detecting whether the waybill to be identified meets preset conditions. When the waybill to be identified meets the preset conditions, an anomaly reason indication information corresponding to the preset conditions is output through the target terminal of the waybill to be identified. When the waybill to be identified does not meet the preset conditions, no anomaly reason indication information is output.

[0199] The target terminal includes at least one of the scanning terminal for the waybill to be identified and the management platform. Furthermore, in some embodiments, the executing entity of the waybill weight information anomaly identification method of this application may also be the target terminal.

[0200] The anomaly cause indication information is used to indicate the cause of the anomaly in the weight information of the waybill to be identified.

[0201] For example, the preset conditions might be that the weight of the waybill to be identified is less than a preset weight threshold and there is no packaging operation on the waybill (e.g., <3kg and no packaging operation), and the anomaly reason indication information is "the weight of the waybill to be identified is less than the preset weight threshold and there is no packaging operation." When it is determined that the weight information of the waybill to be identified is abnormal, it checks whether the waybill to be identified meets the preset conditions. If the weight of the waybill to be identified is less than the preset weight threshold and there is no packaging operation on the waybill to be identified (e.g., <3kg and no packaging operation), the anomaly reason indication information "the weight of the waybill to be identified is less than the preset weight threshold and there is no packaging operation" is output through the target terminal of the waybill to be identified. If the waybill to be identified does not meet the preset conditions, a weight abnormality prompt message or a prompt message suggesting a review of the weight of the waybill to be identified can be output through the target terminal of the waybill to be identified.

[0202] The preset conditions here are for illustrative purposes only. The specific conditions can be adjusted according to the actual business scenario requirements and are not limited to these conditions.

[0203] It is understandable that the weight anomaly alert, the suggestion to review the weight of the waybill to be identified, and the anomaly cause indication can be output as one of them, or all three can be output. Figure 4 This is just an example and is not intended to be limiting.

[0204] For example, the scanning terminal can be a terminal used to scan waybill identification codes during waybill pickup, transfer, and delivery, or it can be a sorting device used during waybill sorting. The management platform can be in the form of a server or a terminal. Here, there is no limitation on the specific form of the scanning terminal and the management platform; they can vary according to the actual business scenario.

[0205] like Figure 5 As shown, Figure 5 This is a schematic diagram illustrating the output of prompt information by the target terminal according to an embodiment of this application. In actual business scenarios, a single target terminal (such as a scanning terminal) can simultaneously output prompt information for abnormal weight, a prompt suggesting verification of the weight of the waybill to be identified, and an indication of the cause of the abnormality, such as... Figure 5 As shown in (a) in the figure, Figure 5 (a) in the text indicates that the scanning terminal simultaneously outputs a weight anomaly prompt, a prompt suggesting verification of the weight of the waybill to be identified, and an anomaly cause indication.

[0206] Alternatively, at least two target terminals can simultaneously output a weight anomaly alert, a suggestion to review the weight of the waybill to be identified, and an indication of the cause of the anomaly, such as... Figure 5 As shown in (a) and (b), Figure 5 (b) indicates that the management platform simultaneously outputs information such as weight anomaly alerts, suggestions to review the weight of the waybill to be identified, and anomaly cause indications.

[0207] Or, such as Figure 6 As shown, Figure 6 This is another schematic diagram of the target terminal outputting prompt information provided in the embodiments of this application. It can also output at least one of the following based on at least two target terminals: weight abnormality prompt information, prompt information suggesting verification of the weight of the waybill to be identified, and abnormality reason indication information. Figure 6 (a) in the text indicates a prompt message suggesting verification of the weight of the waybill to be identified via scanning terminal. Figure 6 (b) indicates that the management platform simultaneously outputs information such as weight anomaly alerts, suggestions to review the weight of the waybill to be identified, and anomaly cause indications.

[0208] To better implement the waybill weight information anomaly identification method in this application embodiment, based on the waybill weight information anomaly identification method, this application embodiment also provides a waybill weight information anomaly identification device, such as... Figure 7 The diagram shown is a structural schematic of one embodiment of the waybill weight information anomaly identification device in this application. The waybill weight information anomaly identification device 700 includes:

[0209] The first acquisition unit 701 is used to acquire target waybill feature information of the waybill to be identified, wherein the target waybill feature information includes multiple waybill information of the waybill to be identified;

[0210] The second acquisition unit 702 is used to acquire the target feature representation value of the waybill to be identified based on the target waybill feature information;

[0211] The identification unit 703 is used to determine the target weight anomaly information of the waybill to be identified based on the target feature representation value and the preset anomaly mapping parameters. The anomaly mapping parameters are used to reflect the constraint relationship between the feature representation value of the waybill and the weight anomaly information of the waybill. The anomaly mapping parameters are learned by learning the sample feature representation value and the actual weight anomaly information of the sample waybill.

[0212] In some embodiments of this application, the target waybill feature information is taken from at least one of the basic feature information of the waybill to be identified, the combined feature information of the basic feature information, and the attribution feature information of the waybill to be identified.

[0213] In some embodiments of this application, the second acquisition unit 702 is specifically used for:

[0214] The target anomaly probability corresponding to the target waybill feature information is obtained from the preset anomaly probabilities corresponding to each waybill feature information;

[0215] The target feature representation value is determined based on the target anomaly probability.

[0216] In some embodiments of this application, the second acquisition unit 702 is specifically used for:

[0217] Based on the target anomaly probability and the preset representation value probability relationship, the target derived representation value of the waybill to be identified is determined, wherein the preset representation value probability relationship is used to indicate the relationship between the target anomaly probability and the target derived representation value;

[0218] The target feature representation value is determined based on the target anomaly probability and the target derived representation value.

[0219] In some embodiments of this application, the target waybill feature information includes basic feature information of the waybill to be identified, combined feature information of the basic feature information, and attribution feature information of the waybill to be identified. The second acquisition unit 702 is specifically used for:

[0220] Obtain the first anomaly probability corresponding to the basic feature information, the second anomaly probability corresponding to the combined feature information, and the third anomaly probability corresponding to the attribution feature information;

[0221] Based on the first anomaly probability, second anomaly probability, third anomaly probability and preset representation value probability mapping relationship shown, the first derived representation value, second derived representation value and third derived representation value of the waybill to be identified are determined;

[0222] Based on the first anomaly probability, the second anomaly probability, the third anomaly probability, the first derived representation value, the second derived representation value, and the third derived representation value, the target feature representation value of the waybill to be identified is determined.

[0223] In some embodiments of this application, the waybill weight information anomaly identification device further includes an extraction unit (not shown in the figure). Before determining the target weight anomaly information of the waybill to be identified based on the target feature representation value and preset anomaly mapping parameters, the extraction unit is specifically used for:

[0224] Obtain a training dataset, wherein the training dataset includes multiple sample waybills, and the sample waybills are labeled with abnormal actual weight information;

[0225] Obtain the sample feature representation value of the sample waybill;

[0226] By using the decision tree to be trained, predictions are made based on the feature representation values ​​of the samples to obtain the predicted weight anomaly information of the sample waybills;

[0227] Based on the predicted weight anomaly information and the actual weight anomaly information, the model parameters of the decision tree to be trained are adjusted until the preset stop training condition is met, and a trained decision tree is obtained.

[0228] Extract the model parameters of the trained decision tree to serve as the anomaly mapping parameters.

[0229] In some embodiments of this application, the waybill weight information anomaly identification device further includes an output unit (not shown in the figure), wherein the target weight anomaly information is used to indicate whether the weight information of the waybill to be identified is abnormal. After determining the target weight anomaly information of the waybill to be identified based on the target feature representation value and the preset anomaly mapping parameters, the output unit is specifically used for:

[0230] When it is determined that the weight information of the waybill to be identified is abnormal, a weight abnormality prompt message is output through the target terminal of the waybill to be identified. The target terminal includes at least one of the scanning terminal of the waybill to be identified and the management platform.

[0231] Alternatively, when it is determined that the weight information of the waybill to be identified is abnormal, a prompt message suggesting that the weight of the waybill to be identified be reviewed is output through the target terminal of the waybill to be identified. The target terminal includes at least one of the scanning terminal of the waybill to be identified and the management platform.

[0232] In some embodiments of this application, the target weight anomaly information is used to indicate whether the weight information of the waybill to be identified is abnormal. After determining the target weight anomaly information of the waybill to be identified based on the target feature representation value and the preset anomaly mapping parameter, the output unit is specifically used for:

[0233] When it is determined that the weight information of the waybill to be identified is abnormal, it is checked whether the waybill to be identified meets the preset conditions;

[0234] When the waybill to be identified meets the preset conditions, the target terminal of the waybill to be identified outputs the abnormal reason indication information corresponding to the preset conditions. The target terminal includes at least one of the scanning terminal of the waybill to be identified and the management platform. The abnormal reason indication information is used to indicate the abnormal reason of the weight information of the waybill to be identified.

[0235] In some embodiments of this application, the preset condition refers to the fact that the waybill to be identified is less than a preset weight threshold and there is no package creation operation for the waybill to be identified, and the abnormal reason indication information is that the waybill to be identified is less than the preset weight threshold and there is no package creation operation for the waybill to be identified.

[0236] In practice, each of the above units can be implemented as an independent entity or can be arbitrarily combined to be implemented as the same or several entities. For the specific implementation of each of the above units, please refer to the previous method embodiments, which will not be repeated here.

[0237] Because the waybill weight information anomaly identification device can perform the functions described in this application, Figures 1 to 6 Corresponding to the steps in the waybill weight information anomaly identification method in any embodiment, this application can achieve the following: Figures 1 to 6 For details on the beneficial effects that the waybill weight information anomaly identification method can achieve in any embodiment, please refer to the preceding description, which will not be repeated here.

[0238] Furthermore, to better implement the waybill weight information anomaly identification method in this application embodiment, based on the waybill weight information anomaly identification method, this application embodiment also provides an electronic device, see reference. Figure 8 , Figure 8 This illustration shows a structural diagram of an electronic device according to an embodiment of this application. Specifically, the electronic device provided in this embodiment includes a processor 801, which executes a computer program stored in a memory 802 to implement, for example... Figures 1 to 6 The steps of the waybill weight information anomaly identification method in any embodiment correspond to the following; or, when the processor 801 executes the computer program stored in the memory 802, it implements the following... Figure 7 The functions of each unit in the corresponding embodiment.

[0239] For example, a computer program may be divided into one or more modules / units, one or more of which are stored in memory 802 and executed by processor 801 to complete the embodiments of this application. One or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in a computer device.

[0240] The electronic device may include, but is not limited to, processor 801 and memory 802. Those skilled in the art will understand that the illustrations are merely examples of an electronic device and do not constitute a limitation on the electronic device. It may include more or fewer components than illustrated, or combine certain components, or different components. For example, the electronic device may also include input / output devices, network access devices, buses, etc., with processor 801, memory 802, input / output devices, and network access devices connected via a bus.

[0241] The processor 801 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the electronic device, connecting various parts of the electronic device through various interfaces and lines.

[0242] The memory 802 can be used to store computer programs and / or modules. The processor 801 implements various functions of the computer device by running or executing the computer programs and / or modules stored in the memory 802 and by calling the data stored in the memory 802. The memory 802 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the electronic device (such as audio data, video data, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, RAM, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0243] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the waybill weight information anomaly identification device, electronic equipment, and its corresponding units described above can be found in the following reference: Figures 1 to 6 The description of the method for identifying abnormal waybill weight information in any embodiment will not be repeated here.

[0244] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be performed by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor.

[0245] Therefore, embodiments of this application provide a computer-readable storage medium storing a plurality of instructions that can be loaded by a processor to execute the present application. Figures 1 to 6 For the steps in the method for identifying abnormal waybill weight information in any embodiment, please refer to the following for specific operations: Figures 1 to 6 The description of the method for identifying abnormal waybill weight information in any embodiment will not be repeated here.

[0246] The computer-readable storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0247] Because of the instructions stored in the computer-readable storage medium, the present application can be executed as described above. Figures 1 to 6 Corresponding to the steps in the waybill weight information anomaly identification method in any embodiment, this application can achieve the following: Figures 1 to 6 For details on the beneficial effects that the waybill weight information anomaly identification method can achieve in any embodiment, please refer to the preceding description, which will not be repeated here.

[0248] The foregoing has provided a detailed description of a method, apparatus, electronic device, and computer-readable storage medium for identifying anomalies in waybill weight information according to embodiments of this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for identifying anomalies in waybill weight information, characterized in that, The method includes: Obtain the target waybill feature information of the waybill to be identified, wherein the target waybill feature information includes multiple waybill information of the waybill to be identified; The target anomaly probability corresponding to the target waybill feature information is obtained from the preset anomaly probabilities corresponding to each waybill feature information; Based on the target anomaly probability and the preset representation value probability relationship, the target derived representation value of the waybill to be identified is determined, wherein the preset representation value probability relationship is used to indicate the relationship between the target anomaly probability and the target derived representation value, and the derived representation value is the extended representation information of the waybill feature information; based on the target anomaly probability and the target derived representation value, the target feature representation value of the waybill to be identified is obtained. Based on the target feature representation value and the preset anomaly mapping parameters, the target weight anomaly information of the waybill to be identified is determined. The anomaly mapping parameters are used to reflect the constraint relationship between the feature representation value of the waybill and the weight anomaly information of the waybill. The anomaly mapping parameters are learned by learning the sample feature representation value and the actual weight anomaly information of the sample waybill.

2. The method for identifying abnormal waybill weight information according to claim 1, characterized in that, The target waybill feature information is taken from at least one of the basic feature information of the waybill to be identified, the combined feature information of the basic feature information, and the attribution feature information of the waybill to be identified.

3. The method for identifying anomalies in waybill weight information according to claim 1, characterized in that, The target waybill feature information includes the basic feature information of the waybill to be identified, the combined feature information of the basic feature information, and the attribution feature information of the waybill to be identified; The step of obtaining the target feature representation value of the waybill to be identified includes: Obtain the first anomaly probability corresponding to the basic feature information, the second anomaly probability corresponding to the combined feature information, and the third anomaly probability corresponding to the attribution feature information; Based on the first anomaly probability, second anomaly probability, third anomaly probability and preset representation value probability mapping relationship shown, the first derived representation value, second derived representation value and third derived representation value of the waybill to be identified are determined; Based on the first anomaly probability, the second anomaly probability, the third anomaly probability, the first derived representation value, the second derived representation value, and the third derived representation value, the target feature representation value of the waybill to be identified is determined.

4. The method for identifying abnormal waybill weight information according to claim 1, characterized in that, Before determining the target weight anomaly information of the waybill to be identified based on the target feature representation value and preset anomaly mapping parameters, the method further includes: Obtain a training dataset, wherein the training dataset includes multiple sample waybills, and the sample waybills are labeled with abnormal actual weight information; Obtain the sample feature representation value of the sample waybill; By using the decision tree to be trained, predictions are made based on the feature representation values ​​of the samples to obtain the predicted weight anomaly information of the sample waybills; Based on the predicted weight anomaly information and the actual weight anomaly information, the model parameters of the decision tree to be trained are adjusted until the preset stop training condition is met, and a trained decision tree is obtained. Extract the model parameters of the trained decision tree to serve as the anomaly mapping parameters.

5. The method for identifying abnormal waybill weight information according to claim 1, characterized in that, The target weight anomaly information is used to indicate whether the weight information of the waybill to be identified is abnormal. After determining the target weight anomaly information of the waybill to be identified based on the target feature representation value and the preset anomaly mapping parameters, the method further includes: When it is determined that the weight information of the waybill to be identified is abnormal, a weight abnormality prompt message is output through the target terminal of the waybill to be identified. The target terminal includes at least one of the scanning terminal of the waybill to be identified and the management platform. Alternatively, when it is determined that the weight information of the waybill to be identified is abnormal, a prompt message suggesting that the weight of the waybill to be identified be reviewed is output through the target terminal of the waybill to be identified. The target terminal includes at least one of the scanning terminal of the waybill to be identified and the management platform.

6. The method for identifying anomalies in waybill weight information according to any one of claims 1 to 5, characterized in that, The target weight anomaly information is used to indicate whether the weight information of the waybill to be identified is abnormal. After determining the target weight anomaly information of the waybill to be identified based on the target feature representation value and the preset anomaly mapping parameters, the method further includes: When it is determined that the weight information of the waybill to be identified is abnormal, it is checked whether the waybill to be identified meets the preset conditions; When the waybill to be identified meets the preset conditions, the target terminal of the waybill to be identified outputs the abnormal reason indication information corresponding to the preset conditions. The target terminal includes at least one of the scanning terminal of the waybill to be identified and the management platform. The abnormal reason indication information is used to indicate the abnormal reason of the weight information of the waybill to be identified.

7. The method for identifying anomalies in waybill weight information according to claim 6, characterized in that, The preset conditions refer to the fact that the waybill to be identified is less than the preset weight threshold and there is no package creation operation for the waybill to be identified. The abnormal reason indication information is that the waybill to be identified is less than the preset weight threshold and there is no package creation operation for the waybill to be identified.

8. A device for identifying abnormal weight information on waybills, characterized in that, The waybill weight information anomaly identification device includes: The first acquisition unit is used to acquire target waybill feature information of the waybill to be identified, wherein the target waybill feature information includes multiple waybill information of the waybill to be identified; The second acquisition unit is configured to: acquire the target anomaly probability corresponding to the target waybill feature information from a preset anomaly probability corresponding to each waybill feature information; determine the target derived representation value of the waybill to be identified based on the target anomaly probability and a preset representation value probability relationship, wherein the preset representation value probability relationship is used to indicate the relationship between the target anomaly probability and the target derived representation value, and the derived representation value is the extended representation information of the waybill feature information; and acquire the target feature representation value of the waybill to be identified based on the target anomaly probability and the target derived representation value. The identification unit is used to determine the target weight anomaly information of the waybill to be identified based on the target feature representation value and the preset anomaly mapping parameters. The anomaly mapping parameters are used to reflect the constraint relationship between the feature representation value of the waybill and the weight anomaly information of the waybill. The anomaly mapping parameters are learned by learning the sample feature representation value and the actual weight anomaly information of the sample waybill.

9. An electronic device, characterized in that, It includes a processor and a memory, wherein the memory stores a computer program, and when the processor calls the computer program in the memory, it executes the waybill weight information anomaly identification method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores a computer program, which is loaded by a processor to execute the steps in the waybill weight information anomaly identification method according to any one of claims 1 to 7.

Images