Freight address recommendation method and device, computer equipment and storage medium

Abstract

The invention discloses a freight address recommendation method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining historical order information and historical click information of a target user; determining an offline feature, the offline feature including a short-term offline probability feature; when the target user performs order placing operation or click operation within a first preset time range, real-time order information and real-time click information are obtained; determining a real-time probability feature based on the real-time order information and the real-time click information; replacing the short-term offline probability feature with the real-time probability feature to obtain a new offline feature, and predicting each candidate poi score based on the new offline feature; and based on each candidate poi score, performing corresponding recommendation to the target user. Through fusion of the real-time features and the offline features and prediction of the candidate poi scores, the proportion of recommended order issuing can be greatly increased, the time for re-site selection of the user can be reduced, the order placing experience of the user can be significantly improved, and the order issuing duration can be shortened.

Description

Technical Field

[0001] The present application relates to the field of freight technology, and in particular to a freight address recommendation method, device, computer equipment and storage medium. Background Art

[0002] With the development of science and Internet technology, e-commerce has developed rapidly. More and more people are accustomed to buying the goods they need online, which has facilitated people's lives.

[0003] Currently, when users place an order, they usually need to fill in the address manually, which results in a long order time and poor order efficiency. Although address information can be recommended when users place an order, it is only based on the last order address or the frequently used order address, and the user needs to re-select and confirm the address, which results in a long order time and is not conducive to increasing the order ratio. Summary of the invention

[0004] Based on this, it is necessary to provide a freight address recommendation method, device, computer equipment and storage medium to address the above technical problems, so as to solve at least one problem existing in the above-mentioned prior art.

[0005] In a first aspect, the embodiment of the present application is implemented as follows: a method for recommending a shipping address is provided, comprising:

[0006] Obtain the target user's historical order information and historical click information;

[0007] Based on the historical order information and the historical click information, determining offline features, the offline features including short-term offline probability features;

[0008] When the target user performs an order placement operation or a click operation within a first preset time range, real-time order information and the real-time click information are acquired;

[0009] Determining a real-time probability feature based on the real-time order information and the real-time click information;

[0010] The short-term offline probability feature is replaced by the real-time probability feature to obtain a new offline feature, and based on the new offline feature, the score of each candidate POI is predicted;

[0011] Based on the score of each candidate POI, a corresponding recommendation is made to the target user.

[0012] In one embodiment, after determining whether the target user performs an order placement operation or a click operation within a preset time range, the process includes:

[0013] If not, predict the score of each candidate POI based on the short-term offline probability feature and the long-term offline probability feature.

[0014] In one embodiment, before predicting the score of each candidate POI, the following steps are included:

[0015] Acquire order information of the target user within a second preset time range, wherein the order information includes a shipping address;

[0016] Preprocessing the shipping address to obtain a target shipping address list;

[0017] Based on the target shipping address list, counting the number of shipping addresses;

[0018] When the number of the shipping addresses is less than or equal to a first preset threshold, all the shipping addresses are taken as candidate POIs.

[0019] In one embodiment, after counting the number of candidate shipping addresses, the following steps are included:

[0020] Based on the target shipping address list, counting the number of orders placed by each shipping address;

[0021] When the number of the shipping addresses is greater than the first preset threshold, and the number of orders is less than the second preset threshold, filtering the shipping addresses;

[0022] The filtered shipping address is used as the candidate POI.

[0023] In one embodiment, the order information includes a delivery address, and the preprocessing of the delivery address to obtain a target delivery address list includes:

[0024] De-duplicate the shipping address and the receiving address to obtain a shipping address list and a receiving address list;

[0025] Determine whether the shipping address in the shipping address list is consistent with the shipping address in the shipping address list;

[0026] If so, the shipping addresses in the shipping address list that are consistent with the shipping addresses in the shipping address list are filtered to obtain the target shipping address list.

[0027] In one embodiment, before predicting the score of each candidate POI, the following steps are included:

[0028] Constructing a sample information set, wherein the sample information set includes offline sample features;

[0029] The offline sample features are input into the prediction model for iterative training until a preset convergence condition is met, thereby obtaining a trained prediction model.

[0030] In one embodiment, the offline feature further includes at least one of a time feature, a sequence feature, and a positioning feature.

[0031] In a second aspect, a shipping address recommendation device is provided, comprising:

[0032] A historical information acquisition unit, used to acquire the target user's historical order information and historical click information;

[0033] An offline feature determination unit, configured to determine offline features based on the historical order information and the historical click information, wherein the offline features include short-term offline probability features and long-term offline probability features;

[0034] A real-time information acquisition unit, configured to acquire real-time order information and the real-time click information when the target user performs an order placement operation or a click operation within a first preset time range;

[0035] A real-time probability feature determination unit, configured to determine a real-time probability feature based on the real-time order information and the real-time click information;

[0036] A prediction unit, used to replace the short-term offline probability feature with the real-time probability feature to obtain a new offline feature, and predict the score of each candidate POI based on the new offline feature;

[0037] The recommendation unit is used to make corresponding recommendations to the target user based on the score of each candidate POI.

[0038] In a third aspect, a computer device is provided, comprising a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, wherein the processor implements the steps of the freight address recommendation method as described above when executing the computer-readable instructions.

[0039] In a fourth aspect, a readable storage medium is provided, wherein the readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the steps of the freight address recommendation method as described above are implemented.

[0040] The above-mentioned freight address recommendation method, device, computer equipment and storage medium, the method implementation includes: obtaining the historical order information and historical click information of the target user; based on the historical order information and the historical click information, determining offline features, the offline features include short-term offline probability features and long-term offline probability features; when the target user places an order or clicks within the first preset time range, obtaining real-time order information and the real-time click information; based on the real-time order information and the real-time click information, determining real-time probability features; replacing the short-term offline probability features with the real-time probability features to obtain new offline features, and based on the new offline features, predicting the score of each candidate POI; based on each candidate POI score, making corresponding recommendations to the target user. In the embodiment of the present application, by integrating real-time features with offline features, predicting candidate POIs, the proportion of recommended orders can be greatly increased, and the delivery address can be supplemented and recalled by screening the shipping address, which can reduce the time for users to re-select addresses, significantly improve the user's ordering experience, and shorten the ordering time. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments of the present application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0042] Figure 1 is a flow chart of a method for recommending a shipping address in one embodiment of the present application;

[0043] Figure 2 This is a schematic diagram of an application environment of a method for recommending a freight address in an embodiment of the present application;

[0044] Figure 3 This is a flow chart of a candidate POI supplement method in one embodiment of the present application;

[0045] Figure 4 is a structural schematic diagram of a shipping address recommendation device in one embodiment of the present application;

[0046] Figure 5 It is a schematic diagram of a computer device in one embodiment of the present application. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0048] In one embodiment, if Figure 1 As shown, a method for recommending a shipping address is provided, such as Figure 2 The present invention provides an implementation environment of a method for recommending a freight address, which may specifically include the following steps:

[0049] In step S110, the historical order information and historical click information of the target user are obtained;

[0050] In an embodiment of the present application, historical order information and historical click information within a period of time in the past, such as three months, can be obtained. The historical order information refers to the order information of the user's order for shopping, which may specifically include information such as the shipping address, the receiving address, the user name, and the user ID. The historical click information refers to the user's click operations on different addresses and products, which may specifically include information such as the number of clicks, the frequency of clicks, and the dwell time of clicks. The above-mentioned historical order information and historical click information can be obtained by detecting and recording user operations through an application installed in advance by the client, or a mini-program, a public account, etc.

[0051] In step S120, based on the historical order information and the historical click information, an offline feature is determined, wherein the offline feature includes a short-term offline probability feature;

[0052] In the embodiment of the present application, the offline feature may specifically include a short-term offline probability feature and a long-term offline probability feature, and may also include any one or any combination of a time feature, a sequence feature, and a positioning feature.

[0053] Among them, the short-term offline probability feature can be specifically understood as the offline probability feature of placing orders or clicking at different POIs (Point of Interest) extracted within a short time range, such as one or any combination of offline probability features within 1 day, offline probability features within 3 days, and offline probability features within 7 days. The specific time range and quantity can be set according to the actual situation.

[0054] The long-term offline probability feature may be an offline probability feature of placing an order or clicking under different POIs extracted within a long-term time range, such as an offline probability feature within 180 days.

[0055] Among them, time features can be understood as features related to time, such as order time, such as holidays, rest days, weekdays, daytime or nighttime, etc.

[0056] Among them, the sequence characteristics can be understood as information such as the rising or falling trend of orders or clicks within a certain period of time.

[0057] Among them, the positioning feature can be understood as the location where the user is when placing an order or clicking an operation, the shipping location, the receiving location and other information.

[0058] In the embodiment of the present application, after obtaining the historical order information and click information, various features in the historical order information and click information can be analyzed and extracted through big data analysis or feature extraction models to obtain the offline features. It should be noted that the offline features are calculated in an offline state, which can be stored in a data table, and updated regularly or in real time, and pushed online so that the online service can read the offline features stored in the data table and use them for subsequent predictions.

[0059] In step S130, when the target user performs an order placement operation or a click operation within a first preset time range, real-time order information and the real-time click information are obtained;

[0060] In the embodiment of the present application, since the storage time limit of real-time order information in the memory database redis is usually 48 hours, and the storage time limit of real-time click information in the memory database redis is usually 24 hours, if the user places an order within 48 hours, or clicks within 24 hours, the real-time order information and real-time click information can be extracted from the memory database redis. It can be understood that by detecting whether there is order information or click information in the memory database redis, it is determined whether the target user has placed an order or clicked within the first preset time range.

[0061] In step S140, a real-time probability feature is determined based on the real-time order information and the real-time click information;

[0062] In an embodiment of the present application, based on the acquired real-time order information and real-time click information, the real-time probability features of the user's order or click at different POIs can be predicted through feature extraction engineering extraction, or probability prediction models, such as ResNet, convolutional neural networks, etc., and this probability feature can be used to replace the short-term offline probability feature calculated offline, that is, the probability feature of the user placing an order at different POIs in the past 1 or 3 days.

[0063] It should be noted that if the target user places an order or clicks within the first preset time range, the short-term offline probability feature is directly used for score prediction.

[0064] In step S150, the short-term offline probability feature is replaced by the real-time probability feature to obtain a new offline feature, and based on the new offline feature, the score of each candidate POI is predicted;

[0065] In the embodiment of the present application, a prediction model, such as ResNet, convolutional neural network, etc., can be constructed, and by collecting historical offline features, the prediction model is iteratively trained to obtain a trained prediction model. Then, the short-term offline probability features in the offline features can be replaced with real-time probability features to obtain new offline features, and then the new offline features are input into the pre-trained prediction model for prediction processing, and the score of each candidate POI can be obtained.

[0066] It should be noted that each candidate POI can be determined and screened according to the user's order information and click information. Specifically, it can include multiple candidate POIs, and different users can correspond to different candidate POIs.

[0067] In step S160, based on the score of each candidate POI, a corresponding recommendation is made to the target user.

[0068] In an embodiment of the present application, a score can be predicted for each candidate POI, and the candidate POIs can be sorted in descending order of the scores to obtain a candidate POI list. Then, the candidate POI can be recommended according to the arrangement order of the candidate POI list to reduce the time for users to reselect a location, which can significantly improve the user's ordering experience and shorten the order placement time.

[0069] In an embodiment of the present application, by predicting the score of each candidate POI, the shipping address can be screened, and the receiving address can be supplemented and recalled based on the screened shipping address, and the shipping address can be merged into the receiving address, thereby eliminating the need for users to reselect the shipping address and the receiving address on their own, reducing ordering time and increasing the recommended order ratio.

[0070] In an embodiment of the present application, a method for recommending a freight address is provided, including: obtaining historical order information and historical click information of a target user; determining offline features based on the historical order information and the historical click information, wherein the offline features include short-term offline probability features and long-term offline probability features; obtaining real-time order information and the real-time click information when the target user places an order or clicks within a first preset time range; determining real-time probability features based on the real-time order information and the real-time click information; using the real-time probability features as the short-term offline probability features, and predicting the scores of each candidate POI based on the real-time probability features and the long-term offline probability features; and making corresponding recommendations to the target user based on the scores of each candidate POI. In an embodiment of the present application, by integrating real-time features with offline features, candidate POIs are predicted, which can greatly increase the proportion of recommended orders, and the delivery address can be supplemented and recalled by screening the shipping address, which can reduce the time for users to reselect a site, significantly improve the user's ordering experience, and shorten the ordering time.

[0071] In an embodiment of the present application, after determining whether the target user performs an order placement operation or a click operation within a preset time range, the method includes:

[0072] If not, predict the score of each candidate POI based on the offline features.

[0073] Optionally, since the real-time order information is usually saved in the memory database Redis for 48 hours, and the real-time click information is usually saved in the memory database Redis for 24 hours, if the user has not placed an order in the last 48 hours or has not performed a click operation within 24 hours, that is, there is no real-time data stored in the memory database Redis, then offline features can be extracted from the data table, and then the offline features can be input into the model to obtain the score of each candidate POI.

[0074] It should be noted that through multiple online implementations, the experimental results show that when there is no real-time order information and / or click information, if the short-term offline probability is not used, the candidate POI scores obtained are not ideal, and the order ratio index will be slightly lower than the control group. After adding the use of offline features, the order ratio index has been significantly improved.

[0075] See also Figure 3 In one embodiment of the present application, before predicting the score of each candidate POI, the following steps are included:

[0076] Acquire order information of the target user within a second preset time range, wherein the order information includes a shipping address;

[0077] Preprocessing the shipping address to obtain a target shipping address list;

[0078] Based on the target shipping address list, counting the number of shipping addresses;

[0079] When the number of the shipping addresses is less than or equal to a first preset threshold, all the shipping addresses are taken as candidate POIs.

[0080] Optionally, order information within a second preset time range, such as the first three months, may be collected, and based on the order information, the shipping address and the delivery address may be extracted, and a shipping address list and a delivery address list may be generated respectively. The generated shipping address list and delivery address list may be pre-processed by filtering, removing, removing outliers and null values, and the target shipping address list and the target delivery address list may be generated respectively. Among them, the target shipping address list is defined as opoid_list, and each element in opoid_list is the shipping place poiid, and the target delivery address list is defined as dpoid_list, and each element in dpoid_list is the delivery place poiid. It can be understood that each user may correspond to a shipping place address list and a delivery address list respectively.

[0081] For the target delivery address obtained, the number of each delivery address can be counted. If the number is less than or equal to the preset threshold, for example, less than or equal to 3, all the delivery addresses can be supplemented as candidate POIs. Since some users are high-frequency users, the number of candidate POIIDs for high-frequency users is relatively large. Therefore, in order to improve the quality of supplementary POIIDs, the number of candidate POIIDs needs to be limited.

[0082] In an embodiment of the present application, after counting the number of candidate shipping addresses, the following steps are included:

[0083] Based on the target shipping address list, counting the number of orders placed by each shipping address;

[0084] When the number of the shipping addresses is greater than the first preset threshold, and the number of orders is less than the second preset threshold, filtering the shipping addresses;

[0085] The filtered shipping address is used as the candidate POI.

[0086] Optionally, if the number of shipping addresses is greater than a first preset threshold, such as greater than 3, the number of orders for each shipping address can be counted. If the number of orders is less than a second preset threshold, such as less than 2, it will be filtered. After the filtering is completed, it can be determined whether the number of filtered shipping addresses is less than 7. If so, the filtered shipping addresses are supplemented as candidate POIs. Otherwise, the shipping address with the least number of orders will be filtered again until the number of shipping addresses is less than 7.

[0087] Furthermore, detailed POI information of each candidate POIID can be obtained, such as address, name, longitude and latitude, name and phone number of users who have placed orders with the POIID in the past, etc. At this point, each shipping address can be scored and sorted as a candidate POI for the delivery location.

[0088] In an embodiment of the present application, the order information includes a delivery address, and the preprocessing of the delivery address to obtain a target delivery address list includes:

[0089] De-duplicate the shipping address and the receiving address to obtain a shipping address list and a receiving address list;

[0090] Determine whether the shipping address in the shipping address list is consistent with the shipping address in the shipping address list;

[0091] If so, the shipping addresses in the shipping address list that are consistent with the shipping addresses in the shipping address list are filtered to obtain the target shipping address list.

[0092] Optionally, order information within a second preset time range, such as the first three months, can be collected, and based on the order information, the shipping address and the delivery address are extracted, and a shipping address list and a delivery address list are generated respectively. For the shipping address list and the delivery address list generated respectively, filtering, removal, removal of abnormal values ​​and null values ​​and other pre-processing can be performed respectively, and a target shipping address list and a target delivery address list are generated respectively. Among them, the target shipping address list is defined as opoid_list, each element in opoid_list is the shipping place poiid, and the target delivery address list is defined as dpoid_list, and each element in dpoid_list is the delivery place poiid. Then it can be determined whether the delivery address in the above-mentioned target shipping address list exists in the target delivery address list, that is, whether the delivery address in the target shipping address list is consistent with the delivery address in the target delivery address list. If so, the consistent shipping address is removed from the target shipping address list. So that the shipping address in the target shipping address list has never appeared in the previous delivery address list, so that the shipping address in the shipping address list is used as a candidate poi to supplement the delivery address.

[0093] In an embodiment of the present application, before predicting the score of each candidate POI, the following steps are included:

[0094] Constructing a sample information set, wherein the sample information set includes offline sample features;

[0095] The offline sample features are input into the prediction model for iterative training until a preset convergence condition is met, thereby obtaining a trained prediction model.

[0096] Optionally, historical order information and historical click information can be collected, and feature extraction can be performed through feature extraction engineering or prediction models to obtain offline features, which can specifically include short-term offline probability features, long-term offline probability features, time features, timing features, and positioning features. Then, corresponding sample data sets, verification data sets, and test data sets can be constructed. The prediction model is iteratively trained using the offline sample features in the sample data set. When its loss value is less than a preset threshold, or the number of iterations reaches a preset number, a trained prediction model can be obtained.

[0097] Then, the trained prediction model can be evaluated and adjusted through the offline test features in the test data set, and the trained model can be verified through the offline verification features in the verification set.

[0098] In the embodiment of the present application, by integrating real-time features with offline features, candidate POIs are predicted, which can improve the sorting effect and significantly increase the proportion of recommended orders. The delivery address can be supplemented and recalled by screening the delivery address, and the screened delivery address can be merged into the delivery address, which can reduce the time for users to reselect a site, significantly improve the user's ordering experience, and shorten the order placement time.

[0099] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0100] In one embodiment, a shipping address recommendation device is provided, and the shipping address recommendation device corresponds to the shipping address recommendation method in the above embodiment. Figure 4 As shown, the shipping address recommendation device includes a historical information acquisition unit 10, an offline feature determination unit 20, a real-time information acquisition unit 30, a real-time probability feature determination unit 40, a first prediction unit 50 and a recommendation unit 60. The functional modules are described in detail as follows:

[0101] The historical information acquisition unit 10 is used to acquire the historical order information and historical click information of the target user.

[0102] An offline feature determination unit 20, configured to determine offline features based on the historical order information and the historical click information, wherein the offline features include short-term offline probability features;

[0103] A real-time information acquisition unit 30, configured to acquire real-time order information and the real-time click information when the target user performs an order placement operation or a click operation within a first preset time range;

[0104] A real-time probability feature determination unit 40, configured to determine a real-time probability feature based on the real-time order information and the real-time click information;

[0105] A first prediction unit 50 is used to replace the short-term offline probability feature with the real-time probability feature to obtain a new offline feature, and predict the score of each candidate POI based on the new offline feature;

[0106] The recommendation unit 60 is used to make corresponding recommendations to the target user based on the score of each candidate POI.

[0107] In one embodiment of the present application, the device further includes a second prediction unit for

[0108] If not, predict the score of each candidate POI based on the offline features.

[0109] In one embodiment of the present application, the device further includes a candidate POI supplement unit, which is used to:

[0110] Acquire order information of the target user within a second preset time range, wherein the order information includes a shipping address;

[0111] Preprocessing the shipping address to obtain a target shipping address list;

[0112] Based on the target shipping address list, counting the number of shipping addresses;

[0113] When the number of the shipping addresses is less than or equal to a first preset threshold, all the shipping addresses are taken as candidate POIs.

[0114] In one embodiment of the present application, the candidate POI supplement unit is further used to:

[0115] Based on the target shipping address list, counting the number of orders placed by each shipping address;

[0116] When the number of the shipping addresses is greater than the first preset threshold, and the number of orders is less than the second preset threshold, filtering the shipping addresses;

[0117] The filtered shipping address is used as the candidate POI.

[0118] In an embodiment of the present application, the order information includes a delivery address, a candidate POI supplement unit, and is also used for:

[0119] De-duplicate the shipping address and the receiving address to obtain a shipping address list and a receiving address list;

[0120] Determine whether the shipping address in the shipping address list is consistent with the shipping address in the shipping address list;

[0121] If so, the shipping addresses in the shipping address list that are consistent with the shipping addresses in the shipping address list are filtered to obtain the target shipping address list.

[0122] In one embodiment of the present application, the device further includes a model training unit, which is used to:

[0123] Constructing a sample information set, wherein the sample information set includes offline sample features;

[0124] The offline sample features are input into the prediction model for iterative training until a preset convergence condition is met, thereby obtaining a trained prediction model.

[0125] In an embodiment of the present application, the offline feature further includes at least one of a time feature, a sequence feature, and a positioning feature.

[0126] In the embodiment of the present application, by integrating real-time features with offline features, candidate POIs are predicted, which can improve the sorting effect and significantly increase the proportion of recommended orders. The delivery address can be supplemented and recalled by screening the delivery address, and the screened delivery address can be merged into the delivery address, which can reduce the time for users to reselect a site, significantly improve the user's ordering experience, and shorten the order placement time.

[0127] For the specific definition of the freight address recommendation device, please refer to the definition of the freight address recommendation method above, which will not be repeated here. Each module in the above-mentioned freight address recommendation device can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0128] In one embodiment, a computer device is provided. The computer device may be a terminal device, and its internal structure diagram may be as follows: Figure 5 As shown. The computer device includes a processor, a memory, and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a readable storage medium. The readable storage medium stores computer-readable instructions. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer-readable instructions are executed by the processor, a method for recommending a shipping address is implemented. The readable storage medium provided in this embodiment includes a non-volatile readable storage medium and a volatile readable storage medium.

[0129] In an embodiment of the present application, a computer device is provided, including a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, and when the processor executes the computer-readable instructions, the steps of the freight address recommendation method as described above are implemented.

[0130] In an embodiment of the application, a readable storage medium is provided, wherein the readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the steps of the freight address recommendation method as described above are implemented.

[0131] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing related hardware through computer-readable instructions, and the computer-readable instructions can be stored in a non-volatile readable storage medium or a volatile readable storage medium. When the computer-readable instructions are executed, they may include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, information library or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double information rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0132] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0133] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for recommending a shipping address, characterized in that: The method comprises: Obtain the target user's historical order information and historical click information; Based on the historical order information and the historical click information, determining offline features, the offline features including short-term offline probability features; When the target user performs an order placement operation or a click operation within a first preset time range, real-time order information and the real-time click information are acquired; Determining a real-time probability feature based on the real-time order information and the real-time click information; The short-term offline probability feature is replaced by the real-time probability feature to obtain a new offline feature, and based on the new offline feature, a score of each candidate POI is predicted; Based on the score of each candidate POI, a corresponding recommendation is made to the target user.

2. The method for recommending a shipping address according to claim 1, wherein: After determining whether the target user performs an order placement operation or a click operation within a preset time range, the method further includes: If not, predict the score of each candidate POI based on the offline features.

3. The method for recommending a shipping address according to claim 1, wherein: Before predicting the score of each candidate POI, the following steps are included: Acquire order information of the target user within a second preset time range, wherein the order information includes a shipping address; Preprocessing the shipping address to obtain a target shipping address list; Based on the target shipping address list, counting the number of shipping addresses; When the number of the shipping addresses is less than or equal to a first preset threshold, all the shipping addresses are taken as candidate POIs.

4. The method for recommending a shipping address according to claim 3, wherein: After counting the number of candidate shipping addresses, including Based on the target shipping address list, counting the number of orders placed by each shipping address; When the number of the shipping addresses is greater than the first preset threshold, and the number of orders is less than the second preset threshold, filtering the shipping addresses; The filtered shipping address is used as the candidate POI.

5. The method for recommending a shipping address according to claim 4, wherein: The order information includes a delivery address, and the preprocessing of the delivery address to obtain a target delivery address list includes: De-duplicate the shipping address and the receiving address to obtain a shipping address list and a receiving address list; Determine whether the shipping address in the shipping address list is consistent with the shipping address in the shipping address list; If so, the shipping addresses in the shipping address list that are consistent with the shipping addresses in the shipping address list are filtered to obtain the target shipping address list.

6. The method for recommending a shipping address according to claim 1, wherein: Before predicting the score of each candidate POI, the following steps are included: Constructing a sample information set, wherein the sample information set includes offline sample features; The offline sample features are input into the prediction model for iterative training until a preset convergence condition is met, thereby obtaining a trained prediction model.

7. The method for recommending a shipping address according to any one of claims 1 to 6, characterized in that: The offline feature also includes at least one of a time feature, a sequence feature and a positioning feature.

8. A shipping address recommendation device, characterized in that: The device comprises: A historical information acquisition unit, used to acquire the target user's historical order information and historical click information; An offline feature determination unit, configured to determine offline features based on the historical order information and the historical click information, wherein the offline features include short-term offline probability features; A real-time information acquisition unit, configured to acquire real-time order information and the real-time click information when the target user performs an order placement operation or a click operation within a first preset time range; A real-time probability feature determination unit, configured to determine a real-time probability feature based on the real-time order information and the real-time click information; A prediction unit, used to replace the short-term offline probability feature with the real-time probability feature to obtain a new offline feature, and predict the score of each candidate POI based on the new offline feature; The recommendation unit is used to make corresponding recommendations to the target user based on the score of each candidate POI.

9. A computer device comprising a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, characterized in that: When the processor executes the computer-readable instructions, the steps of the shipping address recommendation method according to any one of claims 1 to 7 are implemented.

10. A readable storage medium storing computer-readable instructions, characterized in that: When the computer-readable instructions are executed by a processor, the steps of the shipping address recommendation method according to any one of claims 1 to 7 are implemented.

Images