An information detection method and device in a logistics scenario, a storage medium, and an electronic device

Abstract

The application discloses a kind of information detection method, device, storage medium and electronic equipment under logistics scene.Information detection method includes: obtaining information to be detected, the multi-dimensional analysis is carried out to the information to be detected, obtains the multi-dimensional feature corresponding to the information to be detected;Based on the error delivery prediction model trained in advance, the multi-dimensional feature corresponding to the information to be detected is predicted and processed, and the first detection data of the information to be detected is obtained;And / or, based on the outbound call prediction model trained in advance, the multi-dimensional feature corresponding to the information to be detected is predicted and processed, and the second detection data of the information to be detected is obtained;Based on the first detection data and / or the second detection data, the target quality detection result of the information to be detected is determined.The automation detection is carried out to information to be detected, improves the intelligentization and detection efficiency of information detection, replaces artificial detection process, saves the human consumption and time consumption in information detection process.

Description

Technical Field

[0001] This invention relates to the field of smart logistics technology, and in particular to an information detection method, device, storage medium and electronic device in a logistics scenario. Background Technology

[0002] In the logistics industry, pickup and delivery information is extremely important. Correct pickup and delivery information can not only improve delivery accuracy and sales, but also provide customers with better service. Incorrect pickup and delivery information may lead to return, delays or loss during the delivery process, which will seriously affect the sales volume and success rate of the delivery.

[0003] In the process of realizing this invention, it was found that the prior art has at least the following technical problems: Currently, the information on sending and receiving is usually manually checked by employees of postal or express companies. During peak logistics periods or when handling a large number of delivery orders, this may lead to delays due to untimely processing, or there may be omissions in the verification that result in unsuccessful delivery. Summary of the Invention

[0004] This invention provides a method, device, storage medium, and electronic device for information detection in a logistics scenario, so as to realize the quality detection of information to be detected in a logistics scenario and avoid the interference of low-quality information on the logistics process.

[0005] According to one aspect of the present invention, an information detection method in a logistics scenario is provided, comprising:

[0006] Obtain the information to be detected, perform multi-dimensional analysis on the information to be detected, and obtain the multi-dimensional features corresponding to the information to be detected;

[0007] Based on a pre-trained error delivery prediction model, the multi-dimensional features corresponding to the information to be detected are predicted to obtain the first detection data of the information to be detected; and / or, based on a pre-trained outbound call prediction model, the multi-dimensional features corresponding to the information to be detected are predicted to obtain the second detection data of the information to be detected.

[0008] Based on the first detection data and / or the second detection data, the target quality detection result of the information to be detected is determined.

[0009] Optionally, the information to be detected includes address information and logistics object information;

[0010] The multi-dimensional analysis includes one or more of the following: feature analysis of the address format dimension, feature analysis of the administrative region dimension, feature analysis of the road dimension, and feature analysis of the entity dimension of the address information, as well as feature analysis of the logistics object information.

[0011] Optionally, the step of performing multi-dimensional analysis on the information to be detected to obtain multi-dimensional features corresponding to the information to be detected includes:

[0012] Based on at least one analysis item corresponding to each dimension, the information to be detected is analyzed to obtain feature data corresponding to at least one analysis item corresponding to each dimension.

[0013] The multi-dimensional features are formed based on the feature data corresponding to at least one analysis item for each of the multiple dimensions.

[0014] Optionally, before analyzing the information to be detected, one or more of the following may also be included:

[0015] The information to be detected is preprocessed, wherein the preprocessing includes one or more of the following: removing irrelevant characters and character normalization; and at least one level of address component information is identified in the address information.

[0016] Optionally, the first detection data characterizes the degree of delivery deviation corresponding to the information to be detected, and the second detection data characterizes the degree of address error when an outbound call event occurs corresponding to the information to be detected;

[0017] The step of determining the target quality detection result of the information to be detected based on the first detection data and / or the second detection data includes: determining a first quality detection result based on the first detection data, wherein the first quality detection result is negatively correlated with the first detection data; and / or determining a second quality detection result based on the second detection data, wherein the second quality detection result is negatively correlated with the second detection data; and determining the target quality detection result of the information to be detected based on the first quality detection result and / or the second quality detection result.

[0018] Optionally, the training method of the error delivery prediction model includes: acquiring sample information and historical delivery process information corresponding to the sample information, wherein the historical delivery process information includes delivery station information and delivery route information; generating a first label corresponding to the sample information based on the delivery station information and the delivery route information; and training the error delivery prediction model to be trained based on multiple samples and the first label corresponding to each sample until a trained error delivery prediction model is obtained.

[0019] Optionally, generating a first label corresponding to the sample information based on the delivery station information and the delivery route information includes: determining estimated station information based on the address information in the sample information; determining station determination information based on the estimated station information and the delivery station information; determining route determination information based on the delivery route information and the estimated route information corresponding to the address information; and generating a first label based on the station determination information and the route determination information.

[0020] Optionally, the training method of the outbound call prediction model includes: acquiring sample information and historical delivery process information corresponding to the sample information, wherein the historical delivery process information includes delivery route information and outbound call event information; generating a second label corresponding to the sample information based on the delivery route information and outbound call event information; and training the outbound call prediction model to be trained based on multiple samples and the second label corresponding to each sample until a trained outbound call prediction model is obtained.

[0021] Optionally, generating a second tag corresponding to the sample information based on the delivery route information and outbound call event information includes: determining the distance information between at least one level of address component information corresponding to the address information in the sample information and the delivery route information; and determining the second tag based on the outbound call event information and the distance information.

[0022] According to another aspect of the present invention, an information detection device for a logistics scenario is provided, comprising:

[0023] The feature extraction module is used to acquire the information to be detected, perform multi-dimensional analysis on the information to be detected, and obtain the multi-dimensional features corresponding to the information to be detected.

[0024] The detection module is used to perform prediction processing on the multi-dimensional features corresponding to the information to be detected based on a pre-trained error delivery prediction model to obtain the first detection data of the information to be detected; and / or, to perform prediction processing on the multi-dimensional features corresponding to the information to be detected based on a pre-trained outbound call prediction model to obtain the second detection data of the information to be detected.

[0025] The target quality detection result determination module is used to determine the target quality detection result of the information to be detected based on the first detection data and / or the second detection data.

[0026] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0027] At least one processor; and

[0028] A memory communicatively connected to the at least one processor; wherein,

[0029] The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the information detection method in a logistics scenario according to any embodiment of the present invention.

[0030] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions, the computer instructions being configured to cause a processor to execute and implement the information detection method in a logistics scenario as described in any embodiment of the present invention.

[0031] The technical solution of this invention, upon obtaining the information to be detected in a logistics scenario, performs quality detection on the information to be detected based on an error delivery prediction model and / or an outbound call prediction model to obtain the target quality detection result of the information to be detected. This allows for automated detection of the information to be detected after the user inputs it, improving the intelligence and efficiency of information detection, replacing the manual detection process, and saving manpower and time. The quality detection of the information to be detected can prompt the user to correct errors, improving information accuracy and further increasing the success rate of delivery in logistics scenarios, avoiding interference from errors in the delivery process and the ineffective consumption of logistics resources.

[0032] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of the present invention, 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 the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1 This is a flowchart of an information detection method in a logistics scenario provided by an embodiment of the present invention;

[0035] Figure 2 This is a schematic diagram of the structure of an information detection device in a logistics scenario provided by an embodiment of the present invention;

[0036] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0037] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0038] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0039] Figure 1 This is a flowchart of an information detection method in a logistics scenario provided by an embodiment of the present invention. This embodiment is applicable to situations where, when a customer writes delivery information, the delivery information is used as the information to be detected for quality inspection, a target quality inspection result of the information to be detected is obtained, and a modification prompt is issued when the target quality inspection result of the information to be detected is abnormal. This method can be executed by an information detection device in a logistics scenario, which can be implemented in hardware and / or software. This information detection device in a logistics scenario can be configured in a mobile terminal such as a mobile phone or tablet computer, a computer, or a server. Figure 1 As shown, the method includes:

[0040] S110. Obtain the information to be detected, perform multi-dimensional analysis on the information to be detected, and obtain the multi-dimensional features corresponding to the information to be detected.

[0041] S120. Based on a pre-trained error delivery prediction model, the multi-dimensional features corresponding to the information to be detected are predicted to obtain the first detection data of the information to be detected; and / or, based on a pre-trained outbound call prediction model, the multi-dimensional features corresponding to the information to be detected are predicted to obtain the second detection data of the information to be detected.

[0042] S130. Based on the first detection data and / or the second detection data, determine the target quality detection result of the information to be detected.

[0043] In this embodiment, an information collection page is displayed. This page is used to collect delivery and receipt information, which may include recipient information and sender information. One or more of the recipient and sender information are used as the information to be detected. The information to be detected includes address information and logistics object information. Taking recipient information as an example, recipient information includes recipient address information and recipient object information; taking sender information as an example, sender information includes sender address information and sender object information. The logistics object refers to the person or organization that receives and / or sends the package. The logistics object information may be the contact information of the logistics object, and includes recipient object information and / or sender object information.

[0044] Upon detecting user input on the information collection page, the system identifies the information to be tested and performs a quality check. If the target quality check result is normal, a logistics transaction is generated based on the tested information, and the corresponding transaction information is sent to the logistics server for logistics personnel to respond. If the target quality check result is abnormal, a prompt message is generated to inform the user that the input is incorrect and to prompt the user to modify the input. The modified information is then re-tested until the new tested information achieves a normal target quality check result.

[0045] In this embodiment, multi-dimensional features are obtained by performing multi-dimensional analysis on the information to be tested, thereby improving the comprehensiveness of the analysis. The multi-dimensional features of the information to be tested are used to perform quality testing on the information to be tested, providing multi-dimensional features as a basis for the quality testing process and improving the accuracy of quality testing.

[0046] Optionally, before analyzing the information to be detected, the information is preprocessed. This preprocessing includes one or more of the following: removing irrelevant characters and character normalization. Irrelevant characters may include, but are not limited to, spaces, punctuation marks, and special characters. For example, an irrelevant character set is pre-defined, and the information to be detected is matched against this set. Characters that match successfully are identified as irrelevant characters and removed from the information to be detected. The characters in the irrelevant character set are pre-defined according to the preprocessing requirements of the information to be detected. Character normalization includes, but is not limited to, conversion between traditional and simplified Chinese characters, and conversion between full-width and half-width characters. By preprocessing the information to be detected, interfering information is removed, improving the accuracy of quality detection.

[0047] The information to be detected includes address information and logistics object information. Accordingly, the multi-dimensional analysis of the information to be detected includes the analysis of address information and / or the analysis of logistics object information. Before analyzing the address information, the analysis also includes identifying at least one level of address component information within the address information, and subsequently analyzing this at least one level of address component information. Address component information includes, but is not limited to: province / autonomous region information, city information, district / county information, township / street / descriptive area information, village / community information, group / team information, AOI / district / phase information, main road / road segment information, directional terms, house number, POI / building / unit / door information, floor / room number information, etc. Identifying the address component information in the information to be detected facilitates the address information analysis process.

[0048] Optionally, at least one level of address component information in the address information can be identified using a pre-set address entity recognition model. That is, the pre-processed address information is processed by the address entity recognition model to output at least one level of address component information corresponding to the address information. The address entity recognition model can be a neural network model, such as a BERT model.

[0049] The training process of this address entity recognition model can be as follows: obtain an initial sample set, which includes multiple sample address information; by labeling the sample address information, obtain the address component information in the sample address information; and use the address component information in the sample address information as the label corresponding to the sample address information.

[0050] To reduce the number of manually labeled samples and improve the sample data, data augmentation is performed on the initial sample set to obtain extended sample address information. The extended sample address information and the sample address information in the initial sample set form the target sample set. The address entity recognition model is trained based on the target sample set.

[0051] Optionally, the data augmentation process can be as follows: Different entity sets are formed using the address component information in the sample address information. For example, the entity sets may include administrative district datasets, road datasets, and community datasets. Entity word replacement is then performed on the sample address information using the address component information from one or more of these entity sets to obtain expanded sample address information.

[0052] Optionally, the data augmentation process may also include: transforming numeric entities in the sample address information or extended sample address information into Chinese or Arabic numeral types, and / or replacing numeric entities in the sample address information or extended sample address information with other numeric content.

[0053] In the above data augmentation process, spelling errors may also be added to the sample address information or extended sample address information to simulate input errors in the information input process and improve the robustness of the address entity recognition model. Spelling errors may be achieved by replacing one or more address components in the sample address information or extended sample address information with misspelled words. For example, the correct words / phrases in the sample address information or extended sample address information may be replaced with misspelled words that sound the same or have similar forms.

[0054] The address entity recognition model to be trained is iteratively trained based on the target sample set. When the training termination condition is met, the trained address entity recognition model is obtained. Taking the address entity recognition model as a BERT model as an example, during the training process of the address entity recognition model, a CRF layer for sequence labeling is set after the output layer of the BERT model to utilize the dependency relationship between labels to assist the training process.

[0055] The address entity recognition model identifies address components in address information and performs multi-dimensional analysis based on these components and / or logistics object information. For each dimension, the information to be detected is analyzed and processed separately to obtain feature data for that dimension. The feature data from multiple dimensions are then concatenated to obtain multi-dimensional features. Optionally, analysis rules for each dimension are pre-defined, and the information to be detected is analyzed separately based on these rules to obtain feature data for each dimension. Optionally, an analysis model for each dimension is pre-defined, and the information to be detected is input into this model to obtain feature data for each dimension. Optionally, a multi-dimensional analysis model is pre-defined; this model can be a multi-head model, and the multi-dimensional features corresponding to the information to be detected are obtained by inputting the information to be detected into the multi-dimensional analysis model.

[0056] Optionally, the information to be detected is subjected to multi-dimensional analysis to obtain multi-dimensional features corresponding to the information to be detected. This includes: analyzing the information to be detected based on at least one analysis item corresponding to each dimension to obtain feature data corresponding to at least one analysis item corresponding to each dimension; and forming the multi-dimensional features based on the feature data corresponding to at least one analysis item corresponding to multiple dimensions. Each analysis item can correspond to an analysis rule. At least one analysis item corresponding to each dimension can be pre-set. When information to be detected is detected, at least one analysis item corresponding to multiple dimensions is invoked to analyze the information to be detected separately. For example, at least one analysis item corresponding to multiple dimensions is invoked sequentially to analyze the information to be detected in order. Alternatively, at least one analysis item corresponding to multiple dimensions is invoked synchronously to analyze the information to be detected in parallel, thereby accelerating the quality detection efficiency of the information to be detected.

[0057] For any dimension, the information to be detected is analyzed based on at least one analysis item of that dimension, yielding an analysis result corresponding to each analysis item. The analysis results corresponding to at least one analysis item can form the feature data corresponding to that dimension. For example, the analysis results corresponding to at least one analysis item can be arranged into an array or vector in a preset order to obtain the feature data. Optionally, the feature data for each dimension can be in vector form, and multi-dimensional features can be in vector or matrix form. Concatenating feature vectors from multiple dimensions in a preset order yields multi-dimensional features.

[0058] Optionally, the multi-dimensional analysis includes feature analysis of the address format dimension, administrative region dimension, road dimension, and entity dimension of the address information, and / or feature analysis of the logistics object information, one or more of these.

[0059] Feature analysis of the address format dimension of address information can be understood as analyzing the address format of the address information. For example, the analysis items corresponding to the address format dimension of address information are shown in Table 1. Address length can be the byte length in the address information. The address length corresponding to the address information is determined by identifying the number of bytes in the address information, i.e., the analysis result corresponding to g1. Address component information corresponding to core address elements is pre-set. For example, core address elements can be district / county information and AOI / district / phase information. The address component information at multiple levels corresponding to the address information is matched with the aforementioned core address elements respectively. A successful match indicates that the address information includes the aforementioned core address elements, and the analysis result corresponding to g2 can be recorded as 1. A failed match indicates that the address information does not include the aforementioned core address elements, and the analysis result corresponding to g2 can be recorded as 0. Treating each address component information corresponding to the address information as an address element, the number of address elements in the address information can be determined. The byte length of each address element can be used as the length of the address element. If the address length is within a reasonable range, the analysis result corresponding to g5 can be recorded as 1; if the address length is outside a reasonable range, the analysis result corresponding to g5 can be recorded as 0. The analysis results corresponding to g1-g5 are formed into vector form, which serves as the feature data of the address form dimension of the address information.

[0060] Table 1

[0061] g2 Does it contain core address elements? g3 Number of address elements g4 Address element length g5 Is the address length within a reasonable range (2-50 characters)?

[0062] Feature analysis at the administrative region level can be understood as analyzing the address component information corresponding to administrative regions in address information. Specifically, it involves determining whether the address component information corresponding to administrative regions in the address information contains at least one type of error. The analysis items corresponding to the administrative region level can be the analysis rules corresponding to the error types of the address component information corresponding to administrative regions. Optionally, feature analysis at the administrative region level can include, but is not limited to, one or more of the following: missing data analysis, conflict analysis, error analysis, and redundancy analysis. Among them, missing data analysis is used to determine whether there are missing provincial-level administrative divisions, missing municipal-level administrative divisions, missing district / county-level administrative divisions, and missing township-level administrative divisions in the address information. Conflict analysis is used to determine whether there are cascading conflicts or multiple administrative regions in the address information. For example, in the address information "Guangzhou City, Hubei Province", Guangzhou City does not belong to Hubei Province, so there is a cascading conflict; the address information "Beilin District, Xi'an City, Shaanxi Province to Wuhan City, Hubei Province" contains multiple administrative regions. Specifically, this can be achieved by creating an address tree corresponding to each province. The root node of any address tree is the province, municipality, or autonomous region, the child nodes of the province are cities, the child nodes of the city are districts / counties, and so on. The multiple administrative regions in the address information are matched in the above address tree. If a path of a provincial administrative region node can be matched in any address tree and the path is unique, it indicates that there is no administrative region conflict in the address information.

[0063] Error analysis is used to determine whether the address information includes merged or abolished administrative regions, whether there are typos in the administrative regions listed in the address information, and whether the administrative region in the address information actually exists. A list of expired administrative regions is pre-set. The administrative region in the address information is matched against this list. If a match is found, the address information contains an incorrect administrative region. If no match is found, the address information does not contain any expired administrative regions. Optionally, a pre-created address tree is configured with nodes containing expired administrative region information. During the matching process between the address information and the address tree, if an expired administrative region is matched, the address information contains an incorrect administrative region. Furthermore, if the administrative region in the address information cannot be matched successfully in the address tree, but can be matched successfully by replacing the administrative region with homophones or similar-looking characters, then the address information contains a typo for the administrative region, indicating that the administrative region in the address information is incorrect. If the administrative region, its homophones, or similar-looking characters in the address information cannot be matched successfully in the address tree, then it is determined that the administrative region in the address information does not exist, meaning that there is an error in the administrative region in the address information.

[0064] Redundancy analysis is used to determine whether there are duplicate administrative regions in address information. If duplicate administrative regions exist, then the address information is considered to have administrative region redundancy. For example, in the address information "Shaanxi Xi'an City, Shaanxi Province, Xi'an City, Beilin District", "Shaanxi Province" and "Xi'an City" are duplicated. Specifically, this can be determined by matching the administrative regions in the address information against an address tree. If two or more administrative region information matches the same node in the address tree, then the address information is considered to have administrative region redundancy.

[0065] Optionally, the analysis items corresponding to the administrative region dimension are shown in Table 2. By determining the analysis results corresponding to each analysis item, the analysis results corresponding to d1-d9 are respectively formed into feature data corresponding to the administrative region dimension.

[0066] Table 2

[0067] d2 There is a lack of municipal-level administrative divisions. d3 There is a lack of district and county-level administrative divisions. d4 There is a lack of township-level administrative divisions. d5 There is an administrative conflict d6 Administrative region invalid d7 Administrative region misspelling d8 The administrative region does not exist. d9 Redundancy in administrative regions

[0068] Road-level feature analysis can be understood as analyzing road information within address information. This analysis may include, but is not limited to, road missing information analysis, road number missing information analysis, road quantity analysis, and road information validity analysis. Road missing information analysis determines whether road information exists in the address information; road number missing information analysis determines whether the road information in the address information includes road numbers; road quantity analysis determines the number of roads included in the address information; and road information validity analysis determines whether the road information in the address information is in a pre-defined road database and whether the road information in the address information matches the administrative region specified in the address information.

[0069] Optionally, the analysis items corresponding to the road dimension are shown in Table 3. By determining the analysis results corresponding to each analysis item, the analysis results corresponding to r1-r5 are respectively formed into feature data corresponding to the road dimension.

[0070] Table 3

[0071] r2 Does the road number exist? r3 Number of roads r4 Are the road names in the database? r5 Does the road belong to the administrative region?

[0072] Entity-level feature analysis can be understood as analyzing entity information within address information. Specifically, it determines whether the address information can be identified as belonging to a specific household by analyzing the entity information within the address information. For example, the analysis items corresponding to the entity dimension are shown in Table 4. By determining the analysis result corresponding to each analysis item, the analysis results corresponding to e1-e11 are used to form the entity-level feature data. The location element in e1 is pre-set address component information.

[0073] Table 4

[0074] e2 Does a road + road number combination exist? e3 Does a combination of village + village number exist? e4 Does an AOI exist, such as the cell name? e5 Does POI exist? e6 Does the building exist? e7 Does a unit exist? e8 Does the floor exist? e9 Does the household room exist? e10 number of poi e11 Number of administrative villages

[0075] Feature analysis of logistics object information can be understood as analyzing the contact information of the logistics object to determine whether it is possible to contact the logistics object through the logistics object information and complete the delivery. Among them, redundant information can be additional words or remarks in addition to telephone number and name, such as "collected by security guard" or "collected by Cainiao Station".

[0076] For example, the analysis items corresponding to the feature analysis of logistics object information are shown in Table 5. The logistics object information (i.e., the non-address information in the information to be detected) is analyzed sequentially through the following multiple analysis items, and the analysis results corresponding to m1-m4 are respectively formed into feature data corresponding to the logistics object information.

[0077] Table 5

[0078] m2 Does the telephone exist? m3 Is there redundant information? m4 Redundant information length

[0079] The feature data corresponding to the above multiple dimensions are concatenated to obtain multi-dimensional features, and quality detection is performed based on these multi-dimensional features. In this embodiment, a machine learning model for quality detection of the information to be detected is pre-set. The machine learning model performs quality detection on the multi-dimensional features to obtain the quality detection result of the information to be detected.

[0080] Optionally, the machine learning model for quality inspection of the information to be inspected includes an error delivery prediction model and / or an outbound call prediction model. The error delivery prediction model predicts error delivery events caused by the information to be inspected, while the outbound call prediction model predicts outbound call events during the delivery process corresponding to the information to be inspected. An error delivery event is an event where an error occurs at the delivery station during the delivery process based on the information to be inspected, and an outbound call event is an event where the correct address needs to be determined through outbound calls during the delivery process based on the information to be inspected.

[0081] The first detection data is obtained by processing multi-dimensional features through an error delivery prediction model. Optionally, the first detection data represents the probability of an error delivery event occurring. The larger the first detection data, the greater the probability of an error delivery event, indicating a lower quality of the information to be detected. Optionally, the first detection data also represents the degree of delivery deviation corresponding to the information to be detected. The degree of delivery deviation corresponding to the information to be detected can represent the deviation distance between the erroneous delivery station and the correct delivery station during the delivery process based on the information to be detected. The deviation distance between the erroneous delivery station and the correct delivery station is positively correlated with the degree of delivery deviation. The first detection data can be data between 0 and 1. A first detection data of zero indicates that no error delivery event will occur during the delivery process based on the information to be detected. A first detection data greater than zero indicates that an error delivery event will occur during the delivery process based on the information to be detected. The larger the first detection data, the greater the deviation distance between the erroneous delivery station and the correct delivery station, and correspondingly, the lower the quality of the information to be detected.

[0082] The second detection data is obtained by processing multi-dimensional features through an outbound call prediction model. Optionally, the second detection data represents the probability of an outbound call event occurring. The larger the second detection data, the greater the probability of an outbound call event occurring during the delivery process based on the data to be detected, i.e., the worse the quality of the information to be detected. Optionally, the second detection data represents the degree of address error when an outbound call event occurs corresponding to the information to be detected. The degree of address error when an outbound call event occurs can be understood as the degree of deviation between the delivery route and the information to be detected during the delivery process based on the address information to be detected. The second detection data can be data between 0 and 1. A second detection data of zero indicates that the delivery will be successful as long as no outbound call event occurs during the delivery process based on the information to be detected. A second detection data greater than zero indicates that an outbound call event will occur during the delivery process based on the information to be detected. The larger the second detection data, the greater the deviation distance between the information to be detected and the actual delivery route, and correspondingly, the worse the quality of the information to be detected.

[0083] In this embodiment, the first detection data and / or the second detection data obtained by predicting the error delivery events and / or outbound call events of the information to be detected during the delivery process through the error delivery prediction model and / or outbound call prediction model can be used to characterize whether the delivery process based on the information to be detected is successful and smooth, so as to determine the target quality detection result of the information to be detected.

[0084] In some embodiments, a target quality detection result for the information to be detected can be determined based on first detection data or second detection data. Optionally, a first quality detection result is determined based on the first detection data, and this first quality detection result is used as the target quality detection result for the information to be detected. The first quality detection result is negatively correlated with the first detection data. For example, the first quality detection result can be calculated based on the following formula: Where a is a hyperparameter and y1 is the first detection data.

[0085] Optionally, a second quality inspection result is determined based on the second detection data, and this second quality inspection result is set as the target quality inspection result for the information to be inspected. The second quality inspection result is negatively correlated with the second detection data. For example, the second quality inspection result can be calculated based on the following formula: Where b is a hyperparameter and y2 is the second detection data.

[0086] In some embodiments, determining the target quality detection result of the information to be detected based on the first detection data and / or the second detection data includes: determining a first quality detection result based on the first detection data, wherein the first quality detection result is negatively correlated with the first detection data; determining a second quality detection result based on the second detection data, wherein the second quality detection result is negatively correlated with the second detection data; and determining the target quality detection result of the information to be detected based on the first quality detection result and / or the second quality detection result.

[0087] For example, the average of the first quality inspection result and the second quality inspection result can be determined as the target quality inspection result of the information to be inspected; or, the weighted average of the first quality inspection result and the second quality inspection result can be determined as the target quality inspection result of the information to be inspected. The larger the target quality inspection result, the better the quality of the information to be inspected; the smaller the target quality inspection result, the worse the quality of the information to be inspected.

[0088] If the target quality detection result is less than the preset threshold, a prompt message can be generated to prompt the user to modify the input information to be detected.

[0089] Optionally, the prompt message may include the reason for the anomaly in the information to be detected. Anomaly causes are identified based on multi-dimensional features. Specifically, the analysis results corresponding to each analysis item in the multi-dimensional features are judged to determine whether the analysis results corresponding to the analysis item are abnormal. The analysis items with abnormal analysis results are identified as having an anomaly cause. For example, an anomaly cause can be generated based on the description content corresponding to the analysis item with abnormal analysis results. For instance, when g5 in the multi-dimensional features is 0, it indicates that the address length is outside the reasonable range, and the anomaly cause can be generated as "address length abnormal." When d7 in the multi-dimensional features is 1, it indicates that there is a typo in the administrative region, and the prompt message "administrative region has a typo" can be generated.

[0090] The technical solution of this embodiment, upon obtaining the information to be detected in a logistics scenario, performs quality detection on the information to be detected based on an error delivery prediction model and / or an outbound call prediction model to obtain the target quality detection result of the information to be detected. After the user inputs the information to be detected, the information to be detected can be automatically detected, improving the intelligence and efficiency of information detection, replacing the manual detection process, and saving manpower and time consumption in the information detection process. Through the quality detection of the information to be detected, the user can be prompted to modify the erroneous information, improving the accuracy of the information, further improving the success rate of delivery in the logistics scenario, and avoiding interference from erroneous information to the delivery process and the ineffective consumption of logistics resources.

[0091] Based on the above embodiments, the error delivery prediction model is pre-trained. The training method of the error delivery prediction model includes: acquiring sample information and historical delivery process information corresponding to the sample information, wherein the historical delivery process information includes delivery station information and delivery route information; generating a first label corresponding to the sample information based on the delivery station information and the delivery route information; training the error delivery prediction model to be trained based on multiple samples and the first label corresponding to each sample until a trained error delivery prediction model is obtained.

[0092] The sample information can be the collection and delivery information from historical logistics operations. The historical delivery process information corresponding to the sample information can be the information generated during the delivery process of historical logistics operations. For example, it can include delivery station information and delivery route information. Among them, the delivery station information can be the last station information in the delivery process of historical logistics operations, and the delivery route information is the delivery range information of the delivery personnel of historical logistics operations. Different delivery personnel correspond to different delivery route information. The correspondence between delivery personnel and delivery route information is pre-entered and stored. The corresponding delivery route information can be determined by the identification of the delivery personnel (such as employee number or name).

[0093] Optionally, the system determines whether a misdelivery event occurred during the delivery process of a historical logistics business by using the historical delivery process information corresponding to the sample information, and generates a first label. For example, when a misdelivery event occurs, the first label can be 1, and when no misdelivery event occurs, the first label can be 0. Accordingly, the misdelivery prediction model trained based on the sample information and the first label corresponding to the sample information can be used to predict the probability that a misdelivery event will occur during the delivery process of the information to be detected.

[0094] Optionally, based on the delivery station information and delivery route information in the historical delivery process information corresponding to the sample information, it can be determined whether misdelivery events occurred during the delivery process of historical logistics operations and the degree of delivery deviation corresponding to the misdelivery events. A first label corresponding to the sample information is then set based on the degree of delivery deviation corresponding to the misdelivery events. Correspondingly, the misdelivery prediction model trained based on the sample information and its corresponding first label can be used to predict the degree of delivery deviation when misdelivery events occur during the delivery process of the information to be detected.

[0095] Based on the above embodiments, estimated site information is determined based on the address information in the sample information, and site determination information is determined based on the estimated site information and the delivery site information. The site determination information can characterize whether an erroneous delivery event occurred during the delivery process of a historical logistics operation. For example, if the estimated site information and the delivery site information are the same, it indicates that no erroneous delivery event occurred during the delivery process of a historical logistics operation, and the site determination information can be recorded as 0; if the estimated site information and the delivery site information are different, it indicates that an erroneous delivery event occurred during the delivery process of a historical logistics operation, and the site determination information can be recorded as 1. The estimated site information can be determined by matching address information with site address information, for example, by determining the estimated site information based on the site address information with the highest similarity to the address information, or by determining it through a site prediction model; this is not limited here.

[0096] Based on the delivery route information and the estimated route information corresponding to the address information, route determination information is determined; based on the station determination information and the route determination information, the degree of delivery deviation is determined, and a first label is generated based on the degree of delivery deviation. The estimated route information corresponding to the address information can be determined by matching the address information with multiple routes corresponding to the estimated station information, or by using a route prediction model. Optionally, a delivery prediction model is pre-set, which can be used to predict the estimated station information and estimated route information corresponding to the address information.

[0097] Determine the distance between the delivery route information and the estimated route information. For example, this could be the distance between the center point of the delivery route information and the center point of the estimated route information. Use this distance as the route determination information.

[0098] The degree of delivery deviation is determined based on the station determination information and the road area determination information. Specifically, the degree of delivery deviation can be determined by multiplying the station determination information and the road area determination information. For example, if the station determination information is 0, the degree of delivery deviation is zero; if the station determination information is 1, the degree of delivery deviation is equal to the road area determination information. The larger the road area determination information, the greater the distance between the delivery road area information and the estimated road area information, and the greater the degree of delivery deviation. In some embodiments, the road area determination information can also be normalized, that is, the distance between the delivery road area information and the estimated road area information can be converted into a value between 0 and 1.

[0099] A fault delivery prediction model to be trained is obtained. Multi-dimensional analysis is performed on the aforementioned sample information to obtain multi-dimensional features corresponding to the sample information. These multi-dimensional features are then input into the fault delivery prediction model to be trained, yielding the first training detection data output by the model. A first loss function is generated based on the first training detection data and the first label corresponding to the sample information. The model parameters of the fault delivery prediction model to be trained are then adjusted based on the first loss function. This training process is iteratively executed until the training termination condition is met, resulting in a well-trained fault delivery prediction model.

[0100] Based on the above embodiments, the outbound call prediction model is pre-trained. The training method of the outbound call prediction model includes: acquiring sample information and historical delivery process information corresponding to the sample information, wherein the historical delivery process information includes delivery route information and outbound call event information; generating a second label corresponding to the sample information based on the delivery route information and outbound call event information; training the outbound call prediction model to be trained based on multiple sample information and the second label corresponding to each sample information until a trained outbound call prediction model is obtained.

[0101] Outbound call event information can be understood as an identifier indicating whether an outbound call event has occurred. For example, an identifier of 1 indicates that an outbound call event occurred during the delivery process of the historical logistics business corresponding to the sample information, while an identifier of 0 indicates that no outbound call event occurred during the delivery process of the historical logistics business corresponding to the sample information. Optionally, a second label is generated based on the outbound call event information. If the outbound call event information is 1, the second label can be set to 1; if the outbound call event information is 0, the second label can be set to 0. Accordingly, the trained outbound call prediction model can be used to predict the probability of an outbound call event occurring during the delivery process based on the information to be detected.

[0102] Optionally, generating a second tag corresponding to the sample information based on the delivery route information and outbound call event information includes: determining the distance information between at least one level of address component information corresponding to the address information in the sample information and the delivery route information; and determining the second tag based on the outbound call event information and the distance information.

[0103] The degree of address error is characterized by the distance information between at least one level of address component information corresponding to the address information in the sample information and the delivery route information. The greater the distance information, the greater the degree of address error.

[0104] The address information corresponding to at least one level of address component information may include one or more of the following: administrative region, village, road information, neighborhood, and POI. The distance information between each of these address component information and the delivery route information is determined, and the minimum distance among these multiple distance information is determined as the final distance information. Specifically, the distance information between each address component information and the delivery route information may be the distance between the center point of the address component information and the center point of the delivery route information.

[0105] The second label is determined based on the product of outbound call event information and distance information. For example, when the outbound call event information is 0, the second label is 0, indicating that no outbound call event has occurred; when the outbound call event information is 1, the second label is distance information, indicating that an outbound call event has occurred, as well as the degree of address error.

[0106] A training outbound call prediction model is obtained. The sample information is analyzed in multiple dimensions to obtain multi-dimensional features. These features are then input into the model to generate the second training detection data. A second loss function is generated based on the second training detection data and the second labels corresponding to the sample information. The model parameters are adjusted based on this second loss function. This training process is iteratively executed until the training termination condition is met, resulting in a well-trained outbound call prediction model.

[0107] Figure 2 This is a schematic diagram of the structure of an information detection device in a logistics scenario provided by an embodiment of the present invention. Figure 2 As shown, the device includes:

[0108] The feature extraction module 210 is used to acquire the information to be detected, perform multi-dimensional analysis on the information to be detected, and obtain the multi-dimensional features corresponding to the information to be detected.

[0109] The detection module 220 is used to perform prediction processing on the multi-dimensional features corresponding to the information to be detected based on a pre-trained error delivery prediction model to obtain the first detection data of the information to be detected; and / or, to perform prediction processing on the multi-dimensional features corresponding to the information to be detected based on a pre-trained outbound call prediction model to obtain the second detection data of the information to be detected.

[0110] The target quality detection result determination module 230 is used to determine the target quality detection result of the information to be detected based on the first detection data and / or the second detection data.

[0111] The technical solution of this embodiment, upon obtaining the information to be detected in a logistics scenario, performs quality detection on the information to be detected based on an error delivery prediction model and / or an outbound call prediction model to obtain the target quality detection result of the information to be detected. After the user inputs the information to be detected, the information to be detected can be automatically detected, improving the intelligence and efficiency of information detection, replacing the manual detection process, and saving manpower and time consumption in the information detection process. Through the quality detection of the information to be detected, the user can be prompted to modify the erroneous information, improving the accuracy of the information, further improving the success rate of delivery in the logistics scenario, and avoiding interference from erroneous information to the delivery process and the ineffective consumption of logistics resources.

[0112] Based on the above embodiments, optionally, the information to be detected includes address information and logistics object information; the multi-dimensional analysis includes feature analysis of the address form dimension, administrative region dimension, road dimension and entity dimension of the address information, and feature analysis of the logistics object information, as well as one or more of these.

[0113] Optionally, the feature extraction module 210 is used to analyze the information to be detected based on at least one analysis item corresponding to each dimension, and obtain feature data corresponding to at least one analysis item corresponding to each dimension; and form the multi-dimensional features based on the feature data corresponding to at least one analysis item corresponding to multiple dimensions.

[0114] Based on the above embodiments, optionally, the feature extraction module 210 is used to perform one or more of the following: preprocessing the information to be detected, wherein the preprocessing includes one or more of removing irrelevant characters and character normalization processing; and identifying at least one level of address component information in the address information.

[0115] Based on the above embodiments, optionally, the first detection data characterizes the degree of delivery deviation corresponding to the information to be detected, and the second detection data characterizes the degree of address error when an outbound call event occurs corresponding to the information to be detected;

[0116] The target quality inspection result determination module 230 is used to: determine a first quality inspection result based on the first inspection data, wherein the first quality inspection result is negatively correlated with the first inspection data; and / or, determine a second quality inspection result based on the second inspection data, wherein the second quality inspection result is negatively correlated with the second inspection data; and determine the target quality inspection result of the information to be inspected based on the first quality inspection result and / or the second quality inspection result.

[0117] Optionally, based on the above embodiments, the device further includes: a first model training module for: acquiring sample information and historical delivery process information corresponding to the sample information, the historical delivery process information including delivery station information and delivery route information; generating a first label corresponding to the sample information based on the delivery station information and the delivery route information; and training the error delivery prediction model to be trained based on multiple samples and the first label corresponding to each sample until a trained error delivery prediction model is obtained.

[0118] Optionally, the first model training module is further configured to: determine estimated site information based on the address information in the sample information; determine site determination information based on the estimated site information and the delivery site information; determine road area determination information based on the delivery road area information and the estimated road area information corresponding to the address information; and generate a first label based on the site determination information and the road area determination information.

[0119] Optionally, based on the above embodiments, the device further includes: a second model training module for: acquiring sample information and historical delivery process information corresponding to the sample information, the historical delivery process information including delivery route information and outbound call event information; generating a second label corresponding to the sample information based on the delivery route information and outbound call event information; and training the outbound call prediction model to be trained based on multiple samples and the second labels corresponding to each sample until a trained outbound call prediction model is obtained.

[0120] Optionally, the second model training module is further configured to: determine the distance information between at least one level of address component information corresponding to the address information in the sample information and the delivery route information; and determine a second tag based on the outbound call event information and the distance information.

[0121] The information detection device for logistics scenarios provided in this embodiment of the invention can execute the information detection method for logistics scenarios provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0122] Figure 3This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. The electronic device 10 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0123] like Figure 3 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0124] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0125] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as information detection methods in logistics scenarios.

[0126] In some embodiments, the information detection method in a logistics scenario can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the information detection method in a logistics scenario described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform the information detection method in a logistics scenario by any other suitable means (e.g., by means of firmware).

[0127] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0128] Computer programs used to implement the information detection method in a logistics scenario according to the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer programs can be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0129] This invention also provides a computer-readable storage medium storing computer instructions for causing a processor to execute an information detection method in a logistics scenario. The method includes: acquiring information to be detected; performing multi-dimensional analysis on the information to be detected to obtain multi-dimensional features corresponding to the information to be detected; performing prediction processing on the multi-dimensional features corresponding to the information to be detected based on a pre-trained error delivery prediction model to obtain first detection data of the information to be detected; and / or performing prediction processing on the multi-dimensional features corresponding to the information to be detected based on a pre-trained outbound call prediction model to obtain second detection data of the information to be detected; and determining a target quality detection result of the information to be detected based on the first detection data and / or the second detection data.

[0130] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0131] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0132] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0133] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0134] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0135] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. An information detection method in a logistics scenario, characterized in that, include: The process involves acquiring information to be detected, performing multi-dimensional analysis on the information to be detected, and obtaining multi-dimensional features corresponding to the information to be detected; wherein, the information to be detected includes address information and logistics object information. Based on a pre-trained error delivery prediction model, the multi-dimensional features corresponding to the information to be detected are predicted to obtain the first detection data of the information to be detected; and / or, based on a pre-trained outbound call prediction model, the multi-dimensional features corresponding to the information to be detected are predicted to obtain the second detection data of the information to be detected. A first quality test result is determined based on the first test data, wherein the first quality test result is negatively correlated with the first test data; and / or, a second quality test result is determined based on the second test data, wherein the second quality test result is negatively correlated with the second test data; The target quality test result of the information to be tested is determined based on the first quality test result and / or the second quality test result. The training method for the error delivery prediction model includes: Obtain sample information and the corresponding historical delivery process information, wherein the historical delivery process information includes delivery station information and delivery route information; A first label corresponding to the sample information is generated based on the delivery station information and the delivery route information; The error delivery prediction model to be trained is trained based on multiple sample information and the first label corresponding to each sample information until a trained error delivery prediction model is obtained. The training method for the outbound call prediction model includes: Obtain sample information and the corresponding historical delivery process information, including delivery route information and outbound call event information; A second tag corresponding to the sample information is generated based on the delivery route information and outbound call event information; The outbound call prediction model is trained based on multiple sample information and the second label corresponding to each sample information until a trained outbound call prediction model is obtained.

2. The method according to claim 1, characterized in that, The multi-dimensional analysis includes one or more of the following: feature analysis of the address format dimension, feature analysis of the administrative region dimension, feature analysis of the road dimension, and feature analysis of the entity dimension of the address information; and feature analysis of the logistics object information. The step of performing multi-dimensional analysis on the information to be detected to obtain the multi-dimensional features corresponding to the information to be detected includes: Based on at least one analysis item corresponding to each dimension, the information to be detected is analyzed to obtain feature data corresponding to at least one analysis item corresponding to each dimension. The multi-dimensional features are formed based on the feature data corresponding to at least one analysis item for each of the multiple dimensions.

3. The method according to claim 2, characterized in that, Before analyzing the information to be detected, one or more of the following are also included: The information to be detected is preprocessed, wherein the preprocessing includes one or more of the following: removal of irrelevant characters and character normalization. Identify at least one level of address component information in the address information.

4. The method according to claim 1, characterized in that, The first detection data characterizes the degree of delivery deviation corresponding to the information to be detected, and the second detection data characterizes the degree of address error when an outbound call event occurs corresponding to the information to be detected.

5. The method according to claim 1, characterized in that, The step of generating a first label corresponding to the sample information based on the delivery station information and the delivery route information includes: Based on the address information in the sample information, the estimated site information is determined, and based on the estimated site information and the delivery site information, the site determination information is determined. The route determination information is determined based on the estimated route information corresponding to the delivery route information and the address information; A first tag is generated based on the station determination information and the road area determination information.

6. The method according to claim 1, characterized in that, The step of generating a second tag corresponding to the sample information based on the delivery route information and outbound call event information includes: Determine the distance information between at least one level of address component information corresponding to the address information in the sample information and the delivery route information; The second tag is determined based on the outbound call event information and the distance information.

7. An information detection device for a logistics scenario, characterized in that, include: The feature extraction module is used to acquire the information to be detected, perform multi-dimensional analysis on the information to be detected, and obtain the multi-dimensional features corresponding to the information to be detected; wherein, the information to be detected includes address information and logistics object information; The detection module is used to perform prediction processing on the multi-dimensional features corresponding to the information to be detected based on a pre-trained error delivery prediction model to obtain the first detection data of the information to be detected; and / or, to perform prediction processing on the multi-dimensional features corresponding to the information to be detected based on a pre-trained outbound call prediction model to obtain the second detection data of the information to be detected. The target quality detection result determination module is used to determine the target quality detection result of the information to be detected based on the first detection data and / or the second detection data; Specifically, the target quality detection result determination module is used to determine a first quality detection result based on the first detection data, wherein the first quality detection result is negatively correlated with the first detection data; and / or, to determine a second quality detection result based on the second detection data, wherein the second quality detection result is negatively correlated with the second detection data; and to determine the target quality detection result of the information to be detected based on the first quality detection result and / or the second quality detection result. The device further includes: The first model training module is used to acquire sample information and historical delivery process information corresponding to the sample information, the historical delivery process information including delivery station information and delivery route information; generate a first label corresponding to the sample information based on the delivery station information and delivery route information; and train the error delivery prediction model to be trained based on multiple sample information and the first label corresponding to each sample information until a trained error delivery prediction model is obtained. The second model training module is used to acquire sample information and historical delivery process information corresponding to the sample information, the historical delivery process information including delivery route information and outbound call event information; generate a second label corresponding to the sample information based on the delivery route information and outbound call event information; train the outbound call prediction model to be trained based on multiple sample information and the second label corresponding to each sample information until a trained outbound call prediction model is obtained.

8. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the information detection method in the logistics scenario according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that are used to cause a processor to execute the information detection method in the logistics scenario as described in any one of claims 1-6.

Images