Logistics activity risk prediction method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention discloses a logistics activity risk prediction method and device, electronic equipment and a storage medium. The method comprises the steps that logistics activity information is acquired; extracting activity basic feature information and activity statistical feature information of the logistics activity information; performing anomaly recognition on the activity basic feature information and the activity statistical feature information through an activity anomaly recognition model to obtain an activity anomaly recognition result; if the activity abnormity identification result shows that the logistics activity information has participation abnormity, taking the logistics activity information as target logistics activity information to be re-planned; wherein the activity anomaly identification model is obtained by training according to the sample data subjected to anomaly scoring, and the anomaly score is obtained by scoring the sample data through an activity anomaly scoring model. The identification of whether the participation abnormity exists in the logistics activity corresponding to the logistics activity information is realized, so that the participation defect of the activity is overcome, and the user experience is improved.

Description

Technical Field

[0001] Embodiments of the present application relate to the field of computer technology, and specifically to a method, device, electronic device, and storage medium for predicting risks in logistics activities, wherein the storage medium includes a computer-readable storage medium. Background Art

[0002] All walks of life usually use marketing activities to achieve the goals of attracting new customers, promoting activity and increasing revenue. For example, in the logistics industry, marketing activities such as issuing vouchers or coupons are usually used to achieve the corresponding goals.

[0003] However, there will always be some "black industries" that use various cheating methods, tools, and even various "wool-pulling groups" and other artificial means to arbitrage, and event planners sometimes set the scope or conditions of the event incorrectly, resulting in consequences that are difficult to recover. At the same time, these consequences not only affect the normal user experience of the event, but also cause the company's financial and reputation losses, and even affect the company's business strategy development.

[0004] Therefore, when promoting and operating an activity, it is necessary to conduct a risk assessment of the activity to avoid risks in the promotion and operation of the activity and reduce the user experience. Summary of the invention

[0005] The embodiments of the present application provide a logistics activity risk prediction method, device, electronic device and computer-readable storage medium, which can realize risk assessment of activities and improve user experience.

[0006] The present application embodiment provides a method for predicting risks of logistics activities, including:

[0007] Obtain the logistics activity information corresponding to the logistics activity to be tested;

[0008] Extracting basic activity feature information and activity statistical feature information of the logistics activity information, wherein the basic activity feature information is basic attribute information of the logistics activity to be measured, and the activity statistical feature information is the result of statistics of user behavior data collected after the promotion of the logistics activity to be measured;

[0009] Performing abnormality identification on the basic feature information of the activity and the statistical feature information of the activity through an activity abnormality identification model to obtain an activity abnormality identification result;

[0010] If the activity anomaly identification result indicates that there is a participation anomaly in the logistics activity to be tested, the logistics activity information is used as the target logistics activity information to be re-planned;

[0011] The activity anomaly recognition model is trained based on sample data after anomaly scoring, and the anomaly score is obtained by scoring the sample data using the activity anomaly scoring model.

[0012] Accordingly, the embodiment of the present application also provides a logistics activity risk prediction device, including:

[0013] An acquisition module is used to obtain logistics activity information corresponding to the logistics activity to be tested;

[0014] An extraction module, used to extract basic activity feature information and activity statistical feature information of the logistics activity information, wherein the basic activity feature information is basic attribute information of the logistics activity to be measured, and the activity statistical feature information is the result of statistics of user behavior data collected after the promotion of the logistics activity to be measured;

[0015] an identification module, configured to perform anomaly identification on the basic feature information of the activity and the statistical feature information of the activity through an activity anomaly identification model to obtain an activity anomaly identification result, wherein the activity anomaly identification model is obtained by training the sample data after anomaly scoring, and the anomaly score is obtained by scoring the sample data through the activity anomaly scoring model;

[0016] The determination module is used to use the logistics activity information as the target logistics activity information to be re-planned if the activity anomaly identification result indicates that there is a participation anomaly in the logistics activity to be tested.

[0017] Optionally, in some embodiments of the present application, the device further includes a training module for training the activity anomaly recognition model, wherein the training module includes:

[0018] An acquisition unit, used to acquire sample logistics activity information and a sample activity tag corresponding to the sample logistics activity information;

[0019] An extraction unit, used to extract sample activity basic feature information and sample activity statistical feature information from the sample logistics activity information;

[0020] A training unit is used to train an activity anomaly recognition model according to the basic feature information of the sample activity, the statistical feature information of the sample activity and the sample activity label.

[0021] In some embodiments of the present application, the sample activity basic characteristic information includes at least one of activity scope information, activity duration type information, promotion type information, discount type information, promotion holiday type information, or activity level information;

[0022] The statistical characteristic information of sample activities includes the number of participants and growth rate within a preset period, the growth rate and participation proportion of addresses outside the activity scope, the growth rate of the number of virtual accounts participating in the activity, the proportion of virtual accounts participating in the activity, the number of consecutive number mobile phones participating and the growth rate information, the proportion of consecutive number mobile phones, the number of internal personnel participating and the growth rate information, the proportion of internal personnel, the access peak and growth rate information within a preset time period, the number of overseas addresses participating and the growth rate information, the proportion of overseas addresses participating, the number of nighttime participants and the growth rate information, or at least one of the nighttime participation proportion information.

[0023] In some embodiments of the present application, the acquisition unit includes:

[0024] The acquisition subunit is used to obtain sample logistics activity information;

[0025] A scoring subunit, used to score the sample logistics activity information through an activity anomaly scoring model to obtain an anomaly score;

[0026] A determination subunit is used to determine a sample activity label of the sample logistics activity information according to the abnormality score.

[0027] In some embodiments of the present application, the sample activity label includes normal activity and abnormal activity, and the determination subunit is specifically used for:

[0028] Sorting the sample logistics activity information from large to small according to the abnormality scores to obtain a sample logistics activity sorting list;

[0029] Selecting a preset proportion of sample logistics activity information ranked high from the sample logistics activity ranking list as the first sample logistics activity information to be marked;

[0030] Using the sample logistics activity information other than the first sample logistics activity information to be marked as the second sample logistics activity information to be marked;

[0031] Abnormal activities are used as sample activity labels for the first sample logistics activity information to be labeled, and normal activities are used as sample activity labels for the second sample logistics activity information to be labeled.

[0032] In some embodiments of the present application, the device further includes a result determination module, and the result determination module includes:

[0033] A scoring unit, configured to score the logistics activity information by using the activity anomaly scoring model to obtain a logistics activity anomaly score, and obtain a reference recognition result according to the logistics activity anomaly score;

[0034] a weight determination unit, configured to determine a first weight ratio of the logistics activity anomaly score according to an absolute value of a difference between the logistics activity anomaly score and a preset correction score, wherein the first weight ratio is positively correlated with the absolute value;

[0035] A result determination unit, configured to determine a final recognition result from the reference recognition result and the activity abnormality recognition result according to the first weight ratio;

[0036] The modules are determined to include:

[0037] The information determination unit is used to use the logistics activity information as target logistics activity information to be re-planned if the final recognition result indicates that there is a participation anomaly in the logistics activity information.

[0038] In some embodiments of the present application, the result determination unit includes:

[0039] a weight determination subunit, configured to use the difference between the sum of the weights and the first weight ratio as a second weight ratio corresponding to the activity abnormality identification result;

[0040] a first result determination subunit, configured to use the reference recognition result as a final recognition result if the first weight ratio is greater than the second weight ratio;

[0041] The second result determination subunit is configured to use the activity abnormality recognition result as a final recognition result if the first weight ratio is less than or equal to the second weight ratio.

[0042] In a third aspect, an embodiment of the present application further provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and which implements the steps in the above-mentioned logistics activity risk prediction method when the computer program is executed by the processor.

[0043] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps in the above-mentioned logistics activity risk prediction method are implemented.

[0044] In a fifth aspect, the embodiments of the present application further provide a computer program product or a computer program, which includes a computer instruction stored in a computer-readable storage medium. The processor of the computer device reads the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes the method provided in various optional implementations described in the embodiments of the present application.

[0045] The embodiment of the present application obtains the logistics activity information corresponding to the logistics activity to be tested, extracts the basic activity feature information and the activity statistical feature information of the logistics activity information, performs abnormal identification on the basic activity feature information and the activity statistical feature information through the activity abnormality identification model, and obtains the activity abnormality identification result. If the activity abnormality identification result indicates that the logistics activity information has participation abnormality, the logistics activity information is used as the target logistics activity information to be re-planned, wherein the activity abnormality identification model is obtained by training the sample data after abnormality scoring, and the abnormality score is obtained after scoring the sample data by the activity abnormality scoring model. Among them, by extracting the basic activity feature information and the activity statistical feature information corresponding to the logistics activity information, and performing abnormal identification on the basic activity feature information and the activity statistical feature information through the activity abnormality identification model, it is possible to identify whether the logistics activity corresponding to the logistics activity information has participation abnormality, and when it is determined that the logistics activity has participation abnormality, the logistics activity information is used as the target logistics activity information to be re-planned, which can facilitate the activity planner to re-plan and update the logistics activity information, so as to overcome the participation defects of the activity and improve the user experience. Among them, scoring sample data through the activity anomaly scoring model can replace manual scoring, identification and labeling of sample data, thereby improving the efficiency of sample data labeling. Among them, the basic attribute information of the logistics activity to be tested and the statistical results of the user behavior data collected after the promotion of the logistics activity to be tested are used to detect the participation anomaly of the logistics activity to be tested, thereby improving the accuracy of the detection results. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the 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 work.

[0047] Figure 1 It is a scenario diagram of the logistics activity risk prediction method provided in the embodiment of the present application;

[0048] Figure 2 It is a flow chart of a logistics activity risk prediction method provided in an embodiment of the present application;

[0049] Figure 3 is another flow chart of the logistics activity risk prediction method provided in an embodiment of the present application;

[0050] Figure 4 It is a structural schematic diagram of a logistics activity risk prediction device provided in an embodiment of the present application;

[0051] Figure 5It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0052] The following will be combined with the drawings in this application to clearly and completely describe the technical solutions in this application. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments in the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0053] The embodiments of the present application provide a method, device, electronic device and computer-readable storage medium for predicting risks in logistics activities. Specifically, the embodiments of the present application provide a logistics activity risk prediction device suitable for electronic devices, wherein the electronic devices include terminal devices and servers, and the terminal devices include but are not limited to mobile phones, laptops, tablet computers or touch screens; the server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content distribution networks (CDN, Content Delivery Network), and big data and artificial intelligence platforms, etc. The server can be directly or indirectly connected via wired or wireless communication.

[0054] The logistics activity risk prediction method of the embodiment of the present application can be executed by the terminal device alone, or by the server alone, or by the terminal device and the server together. Figure 1 , Figure 1 : is a schematic diagram of a scenario in which a terminal device and a server jointly execute the logistics activity risk prediction method provided in an embodiment of the present application, wherein the specific execution process of the logistics activity risk prediction method executed by the terminal device is as follows:

[0055] The terminal device 10 obtains the logistics activity information of the logistics activity to be tested, and then extracts the basic feature information and the statistical feature information of the logistics activity information, and then sends the basic feature information and the statistical feature information of the activity to the server 11, and the server 11 performs abnormality identification on the basic feature information and the statistical feature information of the activity, specifically:

[0056] After receiving the basic feature information and the statistical feature information of the activity, the server 11 inputs the basic feature information and the statistical feature information of the activity into the activity anomaly recognition model to obtain an activity anomaly recognition result, and then returns the activity anomaly recognition result to the terminal device 10. After receiving the activity anomaly recognition result, the terminal device 10 determines whether the activity anomaly recognition result indicates that the logistics activity to be tested corresponding to the logistics activity information has a participation anomaly. If so, the logistics activity information is used as the target logistics activity information to be re-planned.

[0057] In the embodiment of the present application, the activity anomaly recognition model is trained based on the sample data after anomaly scoring, and the anomaly score of the sample data is obtained by scoring the sample data with the activity anomaly scoring model. In the embodiment of the present application, the activity anomaly recognition model and the activity anomaly scoring model can be the same machine learning model or different machine learning models.

[0058] It can be understood that the basic characteristic information of the activity is the basic attribute information of the logistics activity to be tested, such as attribute information such as activity scope, activity type and activity time; the statistical characteristic information of the activity is the result of statistics of the user behavior data collected after the promotion of the logistics activity to be tested, for example, after the promotion of the logistics activity to be tested, the statistics of the user growth and consumption data within a certain period of time and other result information are obtained.

[0059] Among them, in an embodiment of the present application, the judgment of whether the activity anomaly recognition result indicates that the logistics activity information has a participation anomaly can also be performed by the server. For example, after obtaining the activity anomaly recognition result through the activity anomaly recognition model, the server directly judges whether the activity anomaly recognition result indicates that the logistics activity information has a participation anomaly. Subsequently, when it is judged that the activity anomaly recognition result indicates that the logistics activity information has a participation anomaly, the logistics activity information is directly used as the target logistics activity information to be re-planned, and the target logistics activity information is returned to the terminal device. Subsequently, the user (such as an activity planner) can directly determine whether the logistics activity information corresponding to the target logistics activity information has a participation anomaly based on the target logistics activity information displayed on the visual interface of the terminal device. For example, the target logistics activity information is stored in the candidate set. When the user views the candidate set, it can be known whether the logistics activity information has a participation anomaly.

[0060] Among them, in the embodiment of the present application, the existence of abnormal participation means that during the online promotion process of logistics activities, there are risks such as malicious arbitrage and "fleecing", which affects the participation of normal users and affects the user experience.

[0061] Among them, the logistics activity risk prediction method provided in the embodiment of the present application involves machine learning in the field of artificial intelligence. The embodiment of the present application can realize risk assessment of activities and improve user experience.

[0062] Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science. It attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines so that machines have the functions of perception, reasoning and decision-making. Artificial intelligence technology is a comprehensive discipline that covers a wide range of fields, including both hardware-level technology and software-level technology. Among them, artificial intelligence software technology mainly includes computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0063] Among them, machine learning (ML) is a multi-disciplinary interdisciplinary subject, involving probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and other disciplines. It specializes in studying how computers simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to make computers intelligent. Its applications are spread across all areas of artificial intelligence. Machine learning and deep learning usually include artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and learning by teaching.

[0064] For example, in an embodiment of the present application, the logistics activity information to be evaluated is input into the activity anomaly recognition model to obtain the activity anomaly recognition result, and it is determined based on the activity anomaly recognition result whether to use the logistics activity information as the target logistics activity information to be re-planned.

[0065] It can be understood that the embodiment of the present application extracts the basic feature information and activity statistical feature information corresponding to the logistics activity information, and performs abnormal identification on the basic feature information and activity statistical feature information through the activity abnormality identification model, so as to identify whether there is any participation abnormality in the logistics activity corresponding to the logistics activity information, and when it is determined that there is any participation abnormality in the logistics activity, the logistics activity information is used as the target logistics activity information to be re-planned, which can facilitate the activity planners to re-plan and update the logistics activity information, so as to overcome the participation defects of the activity and improve the user experience. Among them, by scoring the sample data through the activity abnormality scoring model, it is possible to replace manual scoring, identification and labeling of the sample data, thereby improving the efficiency of sample data labeling. Among them, the detection of participation abnormalities in the logistics activity to be tested is performed through the basic attribute information of the logistics activity to be tested, and the statistical results of the user behavior data collected after the promotion of the logistics activity to be tested, thereby improving the accuracy of the detection results.

[0066] It should be noted that the order of description of the following embodiments is not intended to limit the priority order of the embodiments.

[0067] See also Figure 2 , Figure 2 A flow chart of a logistics activity risk prediction method provided in an embodiment of the present application. The specific flow of the logistics activity risk prediction method is as follows:

[0068] 101. Obtain logistics activity information corresponding to the logistics activity to be tested.

[0069] It can be understood that logistics activities are marketing activities involved in the logistics industry, the logistics activities to be tested are logistics activities that need to participate in anomaly detection for the embodiments of the present application, and logistics activity information is marketing activity information of logistics activities, which is the attributes of the logistics activities themselves and the total information generated after the promotion of logistics activities and user participation, including static planning information of logistics activities and user feedback information after the promotion of logistics activities. Among them, static planning information refers to the attribute information of the logistics activities themselves, such as activity scope information, activity duration type information, promotion type information, discount type information, promotion holiday type information or activity level information. User feedback information refers to information generated after users participate in the activity, such as user volume, user address, user group classification information or consumption information.

[0070] Correspondingly, by obtaining logistics activity information, it is convenient to identify risk anomalies in the logistics activity information to obtain risk identification results for the logistics activity information, which can then be conducive to re-planning or improving the logistics activity information based on the risk identification results, so as to improve the comprehensiveness of logistics activity information planning and avoid the risks of participating in loopholes or malicious arbitrage.

[0071] 102. Extract basic activity feature information and activity statistical feature information of the logistics activity information, wherein the basic activity feature information is basic attribute information of the logistics activity to be measured, and the activity statistical feature information is the statistical result of user behavior data collected after the promotion of the logistics activity to be measured.

[0072] Among them, the basic characteristic information of the activity is the static planning information for the logistics activity, which is the attribute information of the activity itself. The statistical characteristic information of the activity is the information obtained after the statistics of the user data generated after the promotion of the logistics activity, for example, the user growth data within a period of time, the regional data of the users participating in the activity, etc.

[0073] Among them, by extracting user feedback data of logistics activities after promotion and obtaining activity statistical feature information through statistical analysis, it is convenient to identify whether there is any participation anomaly in the logistics activities to be tested through analysis of the basic feature information of the activities and the statistical feature information of the activities.

[0074] 103. Perform anomaly recognition on the basic feature information of the activity and the statistical feature information of the activity through an activity anomaly recognition model to obtain an activity anomaly recognition result.

[0075] Among them, in the embodiment of the present application, the activity anomaly recognition model is a machine learning model, which is a trained model. The activity anomaly recognition model can output the corresponding activity anomaly recognition result after inputting the basic feature information of the activity and the statistical feature information of the activity. Compared with only analyzing and identifying the basic feature information of the activity, the embodiment of the present application can improve the accuracy of logistics activity information participating in anomaly recognition by adding statistical feature information of the activity as feature data for analysis by the activity anomaly recognition model.

[0076] Among them, in the embodiment set of the present application, the activity statistical feature information can continuously update data according to the promotion of logistics activities. For example, the activity statistical feature information three days after the promotion of the logistics activity is input into the activity anomaly recognition model, and the activity anomaly recognition results for these three days are output; or the activity statistical feature information one week after the promotion of the logistics activity is input into the activity anomaly recognition model, and the activity anomaly recognition results for this week are output.

[0077] Correspondingly, by obtaining the results of activity anomaly identification in different time periods, it is possible to determine in real time whether there are any participation anomalies in logistics activities, and to facilitate timely adjustment of logistics activity information.

[0078] Among them, the activity anomaly recognition model can be deployed on a server, in the cloud, or directly in a terminal device.

[0079] 104. If the activity anomaly identification result indicates that there is a participation anomaly in the logistics activity to be tested, the logistics activity information is used as the target logistics activity information to be re-planned.

[0080] It should be noted that the activity anomaly identification result in the embodiment of the present application includes normal activity or abnormal activity. When the activity anomaly identification result is normal activity, it indicates that the logistics activity corresponding to the logistics activity information has no defects such as arbitrage and parameter configuration errors in the promotion and operation, that is, the activity anomaly identification result indicates that there is no participation anomaly in the logistics activity to be tested; when the activity anomaly identification result is activity anomaly, it indicates that the logistics activity corresponding to the logistics activity information has defects such as arbitrage and parameter configuration errors in the promotion and operation, which in turn affects the user experience, that is, the activity anomaly identification result indicates that there is participation anomaly in the logistics activity to be tested. Therefore, when the activity anomaly identification result of the logistics activity information is identified as activity anomaly, it indicates that there is participation anomaly in the logistics activity corresponding to the logistics activity information. Therefore, it is necessary to re-plan the logistics activity information to achieve the relevant rights and interests of suppliers when promoting activities, avoid malicious arbitrage, and ensure the participation experience of consumers.

[0081] It should be emphasized that in the embodiment of the present application, the activity anomaly recognition model is trained based on the sample data after anomaly scoring, and the anomaly score corresponding to the sample data can be identified and labeled manually or scored and labeled through a machine learning model. For example, in the embodiment of the present application, the activity anomaly scoring model in machine learning is selected to score the sample data to obtain the anomaly score corresponding to the sample data, and then the sample data is labeled as activity abnormal or activity normal based on the anomaly score.

[0082] Correspondingly, the activity anomaly scoring model is also a trained model, and the activity anomaly scoring model may be of the same type as the activity anomaly identification model or may be different. For example, the activity anomaly scoring model may be trained using an isolation forest (iForest, Isolation Forest) model, and the activity anomaly identification model may be trained using a gradient boosting tree (XGBoost) model. Correspondingly, the types of network models used by the activity anomaly scoring model and the activity anomaly identification model are not limited in the embodiments of the present application. Based on the needs of anomaly identification, the corresponding type of network model may be adaptively selected for training. This is not limited here, that is, the activity anomaly scoring model and the activity anomaly identification model in the embodiments of the present application may be replaced by other anomaly detection models (e.g., machine learning models) or corresponding algorithms.

[0083] It is understandable that the embodiment of the present application uses an activity anomaly identification model to identify anomalies in logistics activity information, realizes the identification of whether there are participation anomalies in logistics activities, and when it is identified that there are participation anomalies in the logistics activities corresponding to the logistics activity information, the logistics activity information is used as the target logistics activity information to be re-planned, which can facilitate the activity planners to re-plan and update the logistics activity information, so as to overcome the participation defects of the activity and improve the user experience. Among them, scoring the sample data through the activity anomaly scoring model can replace manual identification and annotation of the sample data, and improve the efficiency of sample data annotation.

[0084] Among them, since the activity anomaly recognition model is a trained model, before obtaining the logistics activity information, the model is also trained by collecting sample data and sample labels corresponding to the sample data to obtain the activity anomaly recognition model, that is, optionally, in some embodiments of the present application, before the step of "obtaining the logistics activity information corresponding to the logistics activity to be tested", the method also includes training the activity anomaly recognition model, including the following steps:

[0085] Obtaining sample logistics activity information and sample activity tags corresponding to the sample logistics activity information;

[0086] Extracting sample activity basic feature information and sample activity statistical feature information from the sample logistics activity information;

[0087] An activity anomaly recognition model is obtained by training according to the sample activity basic feature information, the sample activity statistical feature information and the sample activity label.

[0088] Among them, the sample logistics activity information corresponds to the activities that were first put on the shelves and then participated by consumers, that is, the activities are activities that were announced or completed in the previous historical time. The sample logistics activity information includes the attribute information of the logistics activity itself and the user data generated by the activity operation, that is, the activities that can obtain sample activity labels according to the results of activity participation. Therefore, the previous sample logistics activity information and its corresponding sample activity labels can be collected to train the model, so that the trained model can predict the corresponding activity anomaly detection results according to the input logistics activity information. Among them, the sample activity label includes a label that the sample activity is normal or the sample activity is abnormal.

[0089] It is understood that in the embodiment of the present application, the basic characteristic information of the sample activity is the basic characteristic information contained in the logistics activity, which is information related to the activity arrangement extracted from the logistics activity information itself, such as attribute information, etc. The attribute information may include the time, location, activity commodity type and other characteristics of the activity. The statistical characteristic information of the sample activity is the information obtained by performing data statistics on the operation results of the logistics activity, for example, statistics on user growth and consumption amount within a period of time.

[0090] Among them, by extracting the basic characteristic information of sample activities and the statistical characteristic information of sample activities of sample logistics activity information, the collection of multi-dimensional characteristic information in the logistics activity information is realized, which helps to improve the accuracy of model training.

[0091] Correspondingly, in an embodiment of the present application, for logistics type activities, types or selection results of characteristic information for the logistics type are provided. For example, for sample logistics activity information, the corresponding sample activity basic characteristic information includes: activity scope information, activity duration type information, promotion type information, discount type information, promotion holiday type information or at least one of activity level information.

[0092] Among them, the activity scope characteristic information includes information such as the whole country or some regions, and some regions may be Guangdong, Henan and other regions; the activity duration type information includes information such as single day, single month, daily cycle, monthly cycle, annual cycle, etc.; the promotion type information includes information such as attracting new customers, promoting activation or promotion, etc.; the discount type information includes information such as discount coupons, cash rebate coupons, no threshold discounts, etc.; the promotion holiday type information includes information such as non-holidays, Spring Festival, Double Eleven, 618, etc.; the activity level information includes information such as level one and level two.

[0093] For the sample logistics activity information, the corresponding sample activity statistical feature information may include: the number of participants and the growth rate information within a preset period, the growth rate and participation proportion information of addresses outside the activity range, the growth rate information of the number of virtual accounts participating in the activity, the proportion information of virtual accounts participating in the activity, the number of consecutive number mobile phones participating and the growth rate information, the proportion information of consecutive number mobile phones, the number of internal personnel participating and the growth rate information, the proportion information of internal personnel, the access peak and growth rate information within a preset time period, the number of overseas addresses participating and the growth rate information, the proportion information of overseas addresses participating, the number of nighttime participants and the growth rate information, or at least one of the nighttime participation proportion information.

[0094] Among them, the preset period can be nearly one hour, nearly three hours, nearly six hours, nearly one day, nearly one week, nearly one month or nearly three months after the logistics activity corresponding to the sample logistics activity information is started.

[0095] The address refers to the user's location, also refers to the IP address of the device. The purpose of collecting the address is to determine whether the user is within the activity range, that is, to determine whether there are users outside the activity range. For example, when the activity range is in Guangdong Province, the device address can be used to determine whether there are users outside Guangdong Province.

[0096] Accordingly, there are certain differences between the risk forecast of logistics activities for the whole country (or the whole network) and the risk forecast of logistics activities for the region. For example, the scope of collected data is different, and the type of statistical data is different. Accordingly, the risk forecast of activities for different regional sizes is adaptively adjusted according to their corresponding characteristics.

[0097] Correspondingly, the sample activity labels of the sample logistics activity information can be obtained by manual annotation, but manual annotation has the problem of low efficiency. Therefore, in the embodiment of the present application, the sample activity labels corresponding to the sample logistics activity information are obtained by scoring through the activity anomaly scoring model, that is, optionally, in some embodiments of the present application, the step of "obtaining the sample logistics activity information and the sample activity labels corresponding to the sample logistics activity information" includes:

[0098] Obtain sample logistics activity information;

[0099] Scoring the sample logistics activity information through an activity anomaly scoring model to obtain an anomaly score;

[0100] A sample activity label of the sample logistics activity information is determined according to the abnormality score.

[0101] It is understandable that the activity anomaly scoring model is a machine learning model and a trained model, which can extract features from sample logistics activity information and score the sample logistics activity information based on the extracted features to obtain an anomaly score. In the embodiment of the present application, the feature information based on the activity anomaly scoring model and the activity anomaly identification model can be the same or different.

[0102] Accordingly, after obtaining the anomaly score for the sample logistics activity information, the sample activity label of the sample logistics activity information can be obtained according to the anomaly score. For example, when the anomaly score is high, the activity anomaly can be used as the sample activity label of the sample logistics activity information; when the anomaly score is low, the activity normal can be used as the sample activity label of the sample logistics activity information, wherein the level of the anomaly score can be determined by comparing with the scoring threshold. For example, if the anomaly score is greater than the scoring threshold, the anomaly score is considered to be high; if the anomaly score is less than the scoring threshold, the anomaly score is considered to be low. Accordingly, the sample logistics activity information with the sample activity label of abnormal activity can be used as a negative sample, and the sample logistics activity information with the sample activity label of normal activity can be used as a positive sample.

[0103] Optionally, in the embodiments of the present application, each sample logistics activity information may be sorted according to the anomaly score, and sample logistics activity information of a certain data may be selected as a negative sample according to the sorting result, that is, optionally, in some embodiments of the present application, the sample activity label includes normal activity and abnormal activity, and the step of "determining the sample activity label of the sample logistics activity information according to the anomaly score" includes:

[0104] Sorting the sample logistics activity information from large to small according to the abnormality scores to obtain a sample logistics activity sorting list;

[0105] Selecting a preset proportion of sample logistics activity information ranked high from the sample logistics activity ranking list as the first sample logistics activity information to be marked;

[0106] Using the sample logistics activity information other than the first sample logistics activity information to be marked as the second sample logistics activity information to be marked;

[0107] Abnormal activities are used as sample activity labels for the first sample logistics activity information to be labeled, and normal activities are used as sample activity labels for the second sample logistics activity information to be labeled.

[0108] Among them, in the embodiment of the present application, the sorting is performed from high to low or from large to small according to the anomaly score. Therefore, a higher ranking indicates a higher or larger anomaly score. By selecting a preset proportion of sample logistics activity information with a higher ranking as the first sample logistics activity information to be marked, and using the first sample logistics activity information to be marked as a negative sample, the selected negative sample meets the characteristics of a higher anomaly score, thereby improving the accuracy of negative sample selection.

[0109] The preset ratio may be predefined according to sample quantity requirements or positive and negative sample ratio requirements. For example, in an embodiment of the present application, the preset ratio may be set to 5%.

[0110] Optionally, since the activity anomaly scoring model itself has the function of scoring the sample logistics activity information for anomalies, the anomaly detection result corresponding to the logistics activity information can also be directly obtained through the output of the activity anomaly scoring model. Accordingly, for the detection result of the activity anomaly scoring model and the activity anomaly identification result corresponding to the activity anomaly identification model, in the embodiment of the present application, one of them can be selected as the final detection result by weight, that is, optionally, in some embodiments of the present application, before the step "if the activity anomaly identification result indicates that the logistics activity information has a participation anomaly, then the logistics activity information is used as the target logistics activity information to be re-planned", the method also includes:

[0111] Scoring the logistics activity information by using the activity anomaly scoring model to obtain a logistics activity anomaly score, and obtaining a reference recognition result according to the logistics activity anomaly score;

[0112] Determine a first weight ratio of the logistics activity anomaly score according to the absolute value of the difference between the logistics activity anomaly score and the preset correction score, wherein the first weight ratio is positively correlated with the absolute value;

[0113] Determining a final recognition result from the reference recognition result and the activity abnormality recognition result according to the first weight ratio;

[0114] If the activity anomaly identification result indicates that the logistics activity information has participation anomalies, the logistics activity information is used as target logistics activity information to be re-planned, including:

[0115] If the final recognition result indicates that there is participation anomaly in the logistics activity information, the logistics activity information is used as target logistics activity information to be re-planned.

[0116] It can be understood that the final recognition result is determined from the reference recognition result and the activity abnormality recognition result by using the first weight ratio, thereby improving the accuracy of the final recognition result.

[0117] Among them, determining the first weight ratio of the logistics activity anomaly score based on the absolute value of the difference between the logistics activity anomaly score and the preset correction score helps to improve the accuracy of determining the first weight ratio, thereby improving the accuracy of the final recognition result.

[0118] Among them, in the embodiment of the present application, the first weight ratio and the absolute value are positively correlated. For example, a larger absolute value indicates that the difference between the logistics activity anomaly score and the preset correction score is larger, which means that the logistics activity anomaly score is more inclined to express a certain result and will not express more than one situation, that is, the logistics activity anomaly score is more expressive and more prominent. For example, when the preset correction score is 0.5 (such as the middle value of the score), if the absolute value of the difference is larger, the logistics activity anomaly score is closer to 0 or 1 (that is, the two extreme values ​​of the anomaly score), that is, the expression is closer to a certain result of identification, that is, it is more inclined to identify the logistics activity information as either normal or abnormal. Accordingly, it can be considered that the credibility of the logistics activity anomaly score is higher.

[0119] Therefore, when the absolute value is large, the first weight ratio can be set larger to increase the weight of the recognition result corresponding to the logistics activity anomaly score as the final recognition result; when the absolute value is small, the first weight ratio can be set smaller to reduce the weight of the recognition result corresponding to the logistics activity anomaly score as the final recognition result.

[0120] Accordingly, after obtaining the first weight ratio for the logistics activity anomaly score, the second weight ratio corresponding to the activity anomaly recognition result can be determined according to the first weight ratio, and the selection result of the final recognition result can be determined according to the comparison between the first weight ratio and the second weight ratio. That is, optionally, in some embodiments of the present application, the step of "determining the final recognition result from the reference recognition result and the activity anomaly recognition result according to the first weight ratio" includes:

[0121] Using the difference between the sum of the weights and the first weight ratio as the second weight ratio corresponding to the abnormal activity recognition result;

[0122] If the first weight ratio is greater than the second weight ratio, taking the reference recognition result as the final recognition result;

[0123] If the first weight ratio is less than or equal to the second weight ratio, the activity abnormality recognition result is used as the final recognition result.

[0124] It can be understood that the sum of the weight ratios is generally 1. Therefore, the sum of the weights in the embodiment of the present application is 1. Correspondingly, the second weight ratio is equal to the difference between 1 and the first weight ratio. If the first weight ratio is greater than the second weight ratio, the reference recognition result is selected as the final recognition result. Conversely, if the first weight ratio is less than or equal to the second weight ratio, the activity abnormality recognition result is selected as the final recognition result.

[0125] It can be understood that the embodiment of the present application obtains basic characteristic information of sample activities and statistical characteristic information of sample activities through the extraction of multi-dimensional sample logistics characteristic information, thereby improving the accuracy of the training of the activity anomaly recognition model, and by using the activity anomaly scoring model to score the sample data, the labeling of the sample data is accelerated, and the training efficiency of the activity anomaly recognition model is improved. When the trained activity anomaly recognition model has a certain accuracy, the accuracy of the activity anomaly recognition results of the logistics activity information can be improved, and the use of the model also improves the efficiency of recognition.

[0126] Optionally, in an embodiment of the present application, the sample activity label corresponding to the sample logistics activity information can also be obtained by manual annotation. For example, the sample activity label of the logistics activity is manually determined based on the operating results of the activity, whether arbitrage occurs, whether there are economic losses such as parameter configuration errors, and so on.

[0127] Correspondingly, the sample logistics activity information that is manually labeled as abnormal activity can also be directly used as negative samples to train the abnormal activity recognition model. For example, see Figure 3 As shown, Figure 3: is another schematic diagram of the process of activity risk prediction provided by an embodiment of the present application, wherein the process of activity risk prediction specifically includes:

[0128] 201. Collecting multiple sample logistics activity information;

[0129] 202. Extracting an original data set corresponding to each sample logistics activity information, where the original data set includes basic characteristic information of the sample activity and statistical characteristic information of the sample activity;

[0130] 203. Obtaining a manual identification label, where the manual identification label is obtained by manually identifying basic feature information of the sample activity and statistical feature information of the sample activity, and determining that the corresponding sample logistics activity is abnormal and then marking it;

[0131] 204. Inputting the sample logistics activity information without labels other than the manual identification labels into the activity anomaly scoring model to obtain an anomaly score;

[0132] Among them, in an embodiment of the present application, the activity anomaly scoring model is obtained by training the isolation forest (iForest, IsolationForest) model. For example, an iForest model is constructed, and the number of trees in the model is set to 2000, the number of samples is set to 256, and the number of samples is the number of unlabeled sample logistics activity information in the training set.

[0133] After training, the iForest model calculates an anomaly score for each sample (i.e., anomaly score), which is defined as follows:

[0134]

[0135] c(n)=2H(n-1)-(2(n-1)) / n

[0136] H(k)=ln(k)+ζ,ζ=0.5772156649

[0137] Where S(x,n) is the anomaly score of sample x in a training set with n samples. c(n) is the expected value of the isolation tree depth h(x) of a dataset consisting of n samples. H(k) is a well-defined calculation formula. h(x) is the path length of the sample point, and ξ is the Euler constant.

[0138] Correspondingly, before the unlabeled sample logistics activity information is input into the activity anomaly scoring model, it is also preprocessed. The preprocessing includes data cleaning and feature engineering. The specific steps of data cleaning are: removing dirty data, deduplication, and filling missing data; the specific steps of feature engineering are: normalization of numerical features, discretization of continuous features, one-hot encoding of categorical features, feature combination, feature correlation calculation and dimensionality reduction, etc.

[0139] 205. Construct a negative sample set based on the sample logistics activity information corresponding to the manual identification label and the sample logistics activity information with anomaly scores in the top 5%;

[0140] 206. Construct a positive sample set according to the sample logistics activity information other than the negative sample set in the plurality of sample logistics activity information;

[0141] 207. Constructing a training set based on the positive sample set and the negative sample set, and performing training to obtain an activity anomaly recognition model;

[0142] Among them, in the embodiment of the present application, the activity anomaly recognition model is trained using the gradient boosting tree (XGBoost) model.

[0143] 208. According to the activity anomaly identification model, anomaly identification is performed on the logistics activity information that is currently planned, completed, promoted and operated to obtain the activity anomaly identification result.

[0144] Among them, by using manually identified negative samples to construct a negative sample set, the accuracy of the negative sample set is improved, and the accuracy of activity anomaly recognition model training is improved.

[0145] The embodiments of the present application can more accurately identify abnormal activities by collecting multi-dimensional data of activities, using an unsupervised algorithm to generate negative samples, and then training a machine learning model based on the obtained data set. By providing early warning of risks, event planners can take timely measures to reduce losses and avoid greater impacts.

[0146] In order to facilitate the better implementation of the logistics activity risk prediction method of the present application, the present application also provides a logistics activity risk prediction device based on the above logistics activity risk prediction method. The meanings of the terms are the same as those in the above logistics activity risk prediction method, and the specific implementation details can refer to the description in the method embodiment.

[0147] See also Figure 4 , Figure 4 : is a schematic diagram of the structure of a logistics activity risk prediction device provided in an embodiment of the present application, wherein the logistics activity risk prediction device can be specifically as follows:

[0148] An acquisition module 301 is used to acquire logistics activity information corresponding to the logistics activity to be tested;

[0149] Extraction module 302, used to extract basic activity feature information and activity statistical feature information of the logistics activity information, wherein the basic activity feature information is basic attribute information of the logistics activity to be measured, and the activity statistical feature information is the result of statistics of user behavior data collected after the promotion of the logistics activity to be measured;

[0150] The identification module 303 is used to perform anomaly identification on the basic feature information of the activity and the statistical feature information of the activity through an activity anomaly identification model to obtain an activity anomaly identification result, wherein the activity anomaly identification model is obtained by training the sample data after anomaly scoring, and the anomaly score is obtained by scoring the sample data through the activity anomaly scoring model;

[0151] The determination module 304 is configured to use the logistics activity information as target logistics activity information to be re-planned if the activity anomaly identification result indicates that there is a participation anomaly in the logistics activity to be tested.

[0152] Optionally, in some embodiments of the present application, the device further includes a training module, and the training module includes:

[0153] An acquisition unit, used to acquire sample logistics activity information and a sample activity tag corresponding to the sample logistics activity information;

[0154] An extraction unit, used to extract sample activity basic feature information and sample activity statistical feature information from the sample logistics activity information;

[0155] A training unit is used to train an activity anomaly recognition model according to the basic feature information of the sample activity, the statistical feature information of the sample activity and the sample activity label.

[0156] In some embodiments of the present application, the sample activity basic characteristic information includes at least one of activity scope information, activity duration type information, promotion type information, discount type information, promotion holiday type information, or activity level information;

[0157] The statistical characteristic information of sample activities includes the number of participants and growth rate within a preset period, the growth rate and participation proportion of addresses outside the activity scope, the growth rate of the number of virtual accounts participating in the activity, the proportion of virtual accounts participating in the activity, the number of consecutive number mobile phones participating and the growth rate information, the proportion of consecutive number mobile phones, the number of internal personnel participating and the growth rate information, the proportion of internal personnel, the access peak and growth rate information within a preset time period, the number of overseas addresses participating and the growth rate information, the proportion of overseas addresses participating, the number of nighttime participants and the growth rate information, or at least one of the nighttime participation proportion information.

[0158] In some embodiments of the present application, the acquisition unit includes:

[0159] The acquisition subunit is used to obtain sample logistics activity information;

[0160] A scoring subunit, used to score the sample logistics activity information through an activity anomaly scoring model to obtain an anomaly score;

[0161] A determination subunit is used to determine a sample activity label of the sample logistics activity information according to the abnormality score.

[0162] In some embodiments of the present application, the sample activity label includes normal activity and abnormal activity, and the determination subunit is specifically used for:

[0163] Sorting the sample logistics activity information from large to small according to the abnormality scores to obtain a sample logistics activity sorting list;

[0164] Selecting a preset proportion of sample logistics activity information ranked high from the sample logistics activity ranking list as the first sample logistics activity information to be marked;

[0165] Using the sample logistics activity information other than the first sample logistics activity information to be marked as the second sample logistics activity information to be marked;

[0166] Abnormal activities are used as sample activity labels for the first sample logistics activity information to be labeled, and normal activities are used as sample activity labels for the second sample logistics activity information to be labeled.

[0167] In some embodiments of the present application, the device further includes a result determination module, and the result determination module includes:

[0168] A scoring unit, configured to score the logistics activity information by using the activity anomaly scoring model to obtain a logistics activity anomaly score, and obtain a reference recognition result according to the logistics activity anomaly score;

[0169] a weight determination unit, configured to determine a first weight ratio of the logistics activity anomaly score according to an absolute value of a difference between the logistics activity anomaly score and a preset correction score, wherein the first weight ratio is positively correlated with the absolute value;

[0170] A result determination unit, configured to determine a final recognition result from the reference recognition result and the activity abnormality recognition result according to the first weight ratio;

[0171] The modules are determined to include:

[0172] An information determination unit is used to use the logistics activity information as target logistics activity information to be re-planned if the final recognition result indicates that there is a participation anomaly in the logistics activity information.

[0173] In some embodiments of the present application, the result determination unit includes:

[0174] a weight determination subunit, configured to use the difference between the sum of the weights and the first weight ratio as a second weight ratio corresponding to the activity abnormality identification result;

[0175] a first result determination subunit, configured to use the reference recognition result as a final recognition result if the first weight ratio is greater than the second weight ratio;

[0176] The second result determination subunit is configured to use the activity abnormality recognition result as a final recognition result if the first weight ratio is less than or equal to the second weight ratio.

[0177] In the embodiment of the present application, the acquisition module 301 first acquires the logistics activity information corresponding to the logistics activity to be tested, and then the extraction module 302 extracts the basic activity feature information and the activity statistical feature information of the logistics activity information. Subsequently, the identification module 303 performs anomaly identification on the basic activity feature information and the activity statistical feature information through an activity anomaly identification model to obtain an activity anomaly identification result, wherein the activity anomaly identification model is trained based on the sample data after anomaly scoring, and the anomaly score is obtained by scoring the sample data with the activity anomaly scoring model. Then, if the activity anomaly identification result indicates that the logistics activity information has participation anomalies, the determination module 304 uses the logistics activity information as the target logistics activity information to be re-planned.

[0178] The basic characteristic information of the activity is the basic attribute information of the logistics activity to be measured, and the statistical characteristic information of the activity is the statistical result of the user behavior data collected after the promotion of the logistics activity to be measured.

[0179] It can be understood that by extracting the basic feature information and activity statistical feature information corresponding to the logistics activity information, and performing abnormal identification on the basic feature information and activity statistical feature information through the activity abnormality identification model, it is possible to identify whether there is any participation abnormality in the logistics activity corresponding to the logistics activity information, and when it is determined that there is any participation abnormality in the logistics activity, the logistics activity information is used as the target logistics activity information to be re-planned, which can facilitate the activity planners to re-plan and update the logistics activity information, so as to overcome the participation defects of the activity and improve the user experience. Among them, scoring the sample data through the activity abnormality scoring model can replace manual scoring, identification and labeling of the sample data, and improve the efficiency of sample data labeling. Among them, the detection of participation abnormalities in the logistics activity to be tested is carried out through the basic attribute information of the logistics activity to be tested, and the statistical results of the user behavior data collected after the promotion of the logistics activity to be tested, so as to improve the accuracy of the detection results.

[0180] In addition, the present application also provides an electronic device, such as Figure 5 As shown, it shows a schematic diagram of the structure of the electronic device involved in this application, specifically:

[0181] The electronic device may include components such as a processor 401 with one or more processing cores, a memory 402 with one or more computer-readable storage media, a power supply 403, and an input unit 404. Those skilled in the art will appreciate that Figure 5 The electronic device structure shown in the figure does not constitute a limitation on the electronic device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0182] The processor 401 is the control center of the electronic device, and uses various interfaces and lines to connect various parts of the entire electronic device. By running or executing software programs and / or modules stored in the memory 402, and calling data stored in the memory 402, the processor 401 performs various functions of the electronic device and processes data, thereby monitoring the electronic device as a whole. Optionally, the processor 401 may include one or more processing cores; preferably, the processor 401 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application programs, and the modem processor mainly processes wireless communications. It is understandable that the above-mentioned modem processor may not be integrated into the processor 401.

[0183] The memory 402 can be used to store software programs and modules. The processor 401 executes various functional applications and data processing by running the software programs and modules stored in the memory 402. The memory 402 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 402 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage devices. Accordingly, the memory 402 may also include a memory controller to provide the processor 401 with access to the memory 402.

[0184] The electronic device also includes a power supply 403 for supplying power to each component. Preferably, the power supply 403 can be logically connected to the processor 401 through a power management system, so as to manage charging, discharging, power consumption and other functions through the power management system. The power supply 403 can also include one or more DC or AC power supplies, recharging systems, power failure detection circuits, power converters or inverters, power status indicators and other arbitrary components.

[0185] The electronic device may further include an input unit 404, which may be used to receive input digital or character information and generate keyboard, mouse, joystick, optical or trackball signal input related to user settings and function control.

[0186] Although not shown, the electronic device may also include a display unit, etc., which will not be described in detail here. Specifically in this embodiment, the processor 401 in the electronic device will load the executable file corresponding to the process of one or more application programs into the memory 402 according to the following instructions, and the processor 401 will run the application program stored in the memory 402, thereby implementing the steps in any one of the logistics activity risk prediction methods provided in the embodiments of the present application.

[0187] The embodiment of the present application obtains logistics activity information of the logistics activity to be tested, extracts basic activity feature information and activity statistical feature information of the logistics activity information, performs anomaly identification on the basic activity feature information and activity statistical feature information through an activity anomaly identification model, and obtains an activity anomaly identification result. If the activity anomaly identification result indicates that there is a participation anomaly in the logistics activity to be tested, the logistics activity information is used as the target logistics activity information to be re-planned, wherein the activity anomaly identification model is trained based on sample data after anomaly scoring, and the anomaly score is obtained by scoring the sample data with the activity anomaly scoring model.

[0188] Among them, by extracting the basic feature information and activity statistical feature information corresponding to the logistics activity information, and performing abnormal identification on the basic feature information and activity statistical feature information through the activity abnormality identification model, it is possible to identify whether there is participation abnormality in the logistics activity corresponding to the logistics activity information, and when it is determined that there is participation abnormality in the logistics activity, the logistics activity information is used as the target logistics activity information to be re-planned, which can facilitate the activity planners to re-plan and update the logistics activity information, so as to overcome the participation defects of the activity and improve the user experience. Among them, by scoring the sample data through the activity abnormality scoring model, it is possible to replace manual scoring, identification and labeling of the sample data, and improve the efficiency of sample data labeling. Among them, the participation abnormality of the logistics activity to be tested is detected through the basic attribute information of the logistics activity to be tested, and the statistical results of the user behavior data collected after the promotion of the logistics activity to be tested, so as to improve the accuracy of the detection results.

[0189] The specific implementation of the above operations can be found in the previous embodiments, which will not be described in detail here.

[0190] A person of ordinary skill in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be completed by instructions, or by controlling related hardware through instructions. The instructions may be stored in a computer-readable storage medium and loaded and executed by a processor.

[0191] To this end, the present application provides a computer-readable storage medium, on which a computer program is stored. The computer program can be loaded by a processor to execute the steps in any logistics activity risk prediction method provided in the present application.

[0192] The specific implementation of the above operations can be found in the previous embodiments, which will not be described in detail here.

[0193] The computer-readable storage medium may include: a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0194] Since the instructions stored in the computer-readable storage medium can execute the steps in any logistics activity risk prediction method provided in the present application, the beneficial effects that can be achieved by any logistics activity risk prediction method provided in the present application can be achieved. Please see the previous embodiments for details and will not be repeated here.

[0195] The above is a detailed introduction to a logistics activity risk prediction method, device, electronic device and computer-readable storage medium provided by the present application. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for technical personnel in this field, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

[0196] It should be noted that in the specific implementation of the present application, it involves activity attribute information of logistics activities pushed by merchants, as well as related data such as consumption data generated by consumers after participating in logistics activities. When the above embodiments of the present application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of relevant data need to comply with relevant laws, regulations and standards of relevant countries and regions.

Claims

1. A method for predicting risks in logistics activities, characterized in that: include: Obtain the logistics activity information corresponding to the logistics activity to be tested; Extracting basic activity feature information and activity statistical feature information of the logistics activity information, wherein the basic activity feature information is basic attribute information of the logistics activity to be measured, and the activity statistical feature information is the result of statistics of user behavior data collected after the promotion of the logistics activity to be measured; Performing abnormality identification on the basic feature information of the activity and the statistical feature information of the activity through an activity abnormality identification model to obtain an activity abnormality identification result; If the activity anomaly identification result indicates that there is a participation anomaly in the logistics activity to be tested, the logistics activity information is used as the target logistics activity information to be re-planned; The activity anomaly recognition model is trained based on sample data after anomaly scoring, and the anomaly score is obtained by scoring the sample data using the activity anomaly scoring model.

2. The method according to claim 1, characterized in that Before obtaining the logistics activity information corresponding to the logistics activity to be tested, the method further includes training the activity anomaly recognition model, including the following steps: Obtaining sample logistics activity information and sample activity tags corresponding to the sample logistics activity information; Extracting sample activity basic feature information and sample activity statistical feature information from the sample logistics activity information; An activity anomaly recognition model is obtained by training according to the sample activity basic feature information, the sample activity statistical feature information and the sample activity label.

3. The method according to claim 2, characterized in that The sample activity basic characteristic information includes at least one of activity scope information, activity duration type information, promotion type information, discount type information, promotion holiday type information or activity level information; The sample activity statistical feature information includes the number of participants and growth rate information within a preset period, the growth rate and participation proportion information of addresses outside the activity scope, the growth rate information of the number of virtual accounts participating in the activity, the proportion information of virtual accounts participating in the activity, the number of consecutive number mobile phones participating and the growth rate information, the proportion information of consecutive number mobile phones, the number of internal personnel participating and the growth rate information, the proportion information of internal personnel, the access peak and growth rate information within a preset time period, the number of overseas addresses participating and the growth rate information, the proportion information of overseas addresses participating, the number of nighttime participants and the growth rate information, or at least one of the nighttime participation proportion information.

4. The method according to claim 2, characterized in that: The obtaining of sample logistics activity information and sample activity labels corresponding to the sample logistics activity information includes: Obtain sample logistics activity information; Scoring the sample logistics activity information through an activity anomaly scoring model to obtain an anomaly score; A sample activity label of the sample logistics activity information is determined according to the abnormality score.

5. The method according to claim 4, characterized in that The sample activity label includes normal activity and abnormal activity, and the sample activity label of the sample logistics activity information is determined according to the abnormality score, including: Sorting the sample logistics activity information from large to small according to the abnormality scores to obtain a sample logistics activity sorting list; Selecting a preset proportion of sample logistics activity information ranked top from the sample logistics activity ranking list as the first sample logistics activity information to be marked; Using the sample logistics activity information other than the first sample logistics activity information to be marked as the second sample logistics activity information to be marked; Abnormal activities are used as sample activity labels for the first sample logistics activity information to be labeled, and normal activities are used as sample activity labels for the second sample logistics activity information to be labeled.

6. The method according to claim 1, characterized in that If the activity anomaly identification result indicates that the logistics activity information has a participation anomaly, before using the logistics activity information as target logistics activity information to be re-planned, the method further includes: Scoring the logistics activity information by using the activity anomaly scoring model to obtain a logistics activity anomaly score, and obtaining a reference recognition result according to the logistics activity anomaly score; Determine a first weight ratio of the logistics activity anomaly score according to the absolute value of the difference between the logistics activity anomaly score and the preset correction score, wherein the first weight ratio is positively correlated with the absolute value; Determining a final recognition result from the reference recognition result and the activity abnormality recognition result according to the first weight ratio; If the activity anomaly identification result indicates that the logistics activity information has participation anomalies, the logistics activity information is used as target logistics activity information to be re-planned, including: If the final recognition result indicates that there is participation anomaly in the logistics activity information, the logistics activity information is used as target logistics activity information to be re-planned.

7. The method according to claim 6, characterized in that The determining a final recognition result from the reference recognition result and the activity abnormality recognition result according to the first weight ratio includes: Using the difference between the sum of the weights and the first weight ratio as the second weight ratio corresponding to the abnormal activity recognition result; If the first weight ratio is greater than the second weight ratio, taking the reference recognition result as the final recognition result; If the first weight ratio is less than or equal to the second weight ratio, the activity abnormality recognition result is used as the final recognition result.

8. A logistics activity risk prediction device, characterized in that: include: An acquisition module is used to obtain logistics activity information corresponding to the logistics activity to be tested; An extraction module, used to extract basic activity feature information and activity statistical feature information of the logistics activity information, wherein the basic activity feature information is basic attribute information of the logistics activity to be measured, and the activity statistical feature information is the result of statistics of user behavior data collected after the promotion of the logistics activity to be measured; an identification module, configured to perform anomaly identification on the basic feature information of the activity and the statistical feature information of the activity through an activity anomaly identification model to obtain an activity anomaly identification result, wherein the activity anomaly identification model is obtained by training the sample data after anomaly scoring, and the anomaly score is obtained by scoring the sample data through the activity anomaly scoring model; The determination module is used to use the logistics activity information as the target logistics activity information to be re-planned if the activity anomaly identification result indicates that there is a participation anomaly in the logistics activity to be tested.

9. An electronic device, characterized in that: It includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the logistics activity risk prediction method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the logistics activity risk prediction method according to any one of claims 1 to 7 are implemented.

Images