Address recommendation method, device, electronic device and computer storage medium
By using positioning and behavioral feature data in takeaway APP for address sorting and screening, and combining with scoring models to optimize address recommendations, the problem of insufficient address recommendation in the existing technology is solved, and a more accurate and personalized address recommendation service is achieved.
Patent Information
- Application Number
- CN202411034285.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2044-07-30
AI Technical Summary
The existing takeaway APP address recommendation service cannot accurately recommend delivery addresses based on users' different ordering habits, resulting in the problem of insufficient recommendations.
By obtaining multiple address records of the target user, coarse sorting and filtering based on the current positioning latitude and longitude, and fine sorting based on the behavioral feature data and the current delivery data input scoring model to generate an address list that is more in line with user needs.
It improves the accuracy and practicality of address recommendations, ensures that the recommended address meets the actual needs and expectations of users and adapts to the personalized needs of users.
Smart Images

Figure CN118568369B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of mobile Internet technology, and in particular to an address recommendation method, device, electronic device and computer storage medium. Background Art
[0002] The popularity of mobile devices and the development of mobile Internet technology have provided many conveniences for modern people’s lives. For example, food delivery apps (Applications) on mobile phones allow users to place orders anytime and anywhere, whether on the way to work or on the way home from get off work. As long as users select or fill in the desired delivery address in the address bar on the order form, they can pick up the food they selected when placing the order at the designated location, making dining more convenient and quick.
[0003] In order to improve the user experience, food delivery apps usually show users the addresses recommended by the system when they submit an order. Existing address recommendation services often simply adopt a recommendation strategy to sort the addresses that users have saved in advance on the food delivery app, and then recommend the sorted address list to the user. For example, the pre-saved addresses are sorted according to the address usage time, and the addresses used by users when placing orders recently are placed at the front of the address list.
[0004] However, different users have different ordering habits. If the same recommendation strategy is used, it is difficult to guarantee that a suitable delivery address is recommended to every user. For example, a user is used to ordering food to be delivered to the company at noon on weekdays and to be delivered to his home at night. If the address is sorted only by the time of use, it is impossible to recommend a suitable delivery address for the user when he orders food at night, resulting in inaccurate address recommendations. Summary of the invention
[0005] The embodiments of the present application provide an address recommendation method, device, electronic device and computer storage medium, which realize accurate address recommendation. The above technical solution is as follows:
[0006] In a first aspect, an embodiment of the present application provides an address recommendation method, including:
[0007] Get multiple address records of the target user;
[0008] Roughly sorting the multiple address records according to the current positioning longitude and latitude of the target user, and filtering the multiple address records based on the result of the rough sorting to obtain a first address list corresponding to at least one filtered first address record;
[0009] Obtaining the target user's behavior feature data and current delivery data, and inputting the behavior feature data, current delivery data, and the first address list into a scoring model, and performing fine sorting on the first address list according to an output result of the scoring model to obtain a second address list; the scoring model is trained by historical order data corresponding to at least one historical order submitted by the target user, and the historical order data includes: the target user's historical behavior feature data, historical delivery data, and historical address records;
[0010] The address information corresponding to the second address list is sent to the terminal corresponding to the target user.
[0011] In a possible implementation, each of the multiple address records includes address latitude and longitude information and address validity information; the address latitude and longitude information is used to represent the address latitude and longitude corresponding to each address record, and the address validity information is used to represent whether the address corresponding to each address record is valid;
[0012] The multiple address records are roughly sorted according to the current location latitude and longitude of the target user, and the multiple address records are filtered based on the result of the rough sorting, including:
[0013] Calculate the latitude and longitude distance between the current location latitude and longitude of the target user and the address latitude and longitude contained in each address record;
[0014] Filter out address records whose longitude and latitude distance is less than or equal to the preset longitude and latitude distance and whose addresses are valid.
[0015] In a possible implementation, each address record in the plurality of address records corresponds to an address index; the first address list is constructed by the address indexes corresponding to each first address record;
[0016] After finely sorting the first address list to obtain the second address list, the method further includes:
[0017] The address information is completed according to the address index in the second address list to obtain the address information corresponding to the second address list.
[0018] In a possible implementation, the behavior feature data, the current delivery data, and the first address list are input into a scoring model, and the first address list is precisely sorted according to the output result of the scoring model to obtain a second address list, including:
[0019] Inputting the behavior characteristic data, the current delivery data and the first address list into the scoring model, and obtaining the address recommendation rate of each first address record corresponding to the first address list output by the scoring model;
[0020] The first address list is sorted in descending order based on the address recommendation rate to obtain a second address list.
[0021] In a possible implementation, the method further includes:
[0022] Input the historical behavior feature data, historical delivery data, and historical address records corresponding to each historical order into the scoring model to obtain the predicted address recommendation rate corresponding to each historical address record output by the scoring model;
[0023] When the proportion of historical orders with a predicted address recommendation rate greater than the preset recommendation rate is less than a first preset ratio threshold, the parameters of the scoring model are adjusted, and the step of inputting the historical behavior feature data, historical delivery data, and historical address records corresponding to each historical order into the scoring model is executed again to obtain the predicted address recommendation rate corresponding to each historical address record output by the scoring model, until the proportion of historical orders with a predicted address recommendation rate greater than the preset recommendation rate is greater than or equal to the first preset ratio threshold.
[0024] Inputting the behavior characteristic data, the current delivery data and the first address list into the scoring model, and obtaining the address recommendation rate of each first address record corresponding to the first address list output by the scoring model;
[0025] The first address list is sorted in descending order based on the address recommendation rate to obtain a second address list.
[0026] In a possible implementation manner, after sending the second address list to a terminal corresponding to the target user, the method further includes:
[0027] The behavior feature data and the current delivery data are input into the prediction model to obtain the address modification rate and the address fulfillment rate corresponding to the current order submitted by the target user; the prediction model includes a first prediction model for predicting the order address modification rate and a second prediction model for predicting the order address fulfillment rate; the first prediction model and the second prediction model are trained by historical order data;
[0028] When the address modification rate corresponding to the current order is greater than the preset modification rate, or the address fulfillment rate corresponding to the current order is less than the preset fulfillment rate, the address input prompt information is sent to the terminal corresponding to the target user;
[0029] When the address modification rate corresponding to the current order is less than or equal to the preset modification rate, or the address fulfillment rate corresponding to the current order is greater than or equal to the preset fulfillment rate, an address confirmation prompt message is sent to the terminal corresponding to the target user.
[0030] In a possible implementation, the method further includes:
[0031] Inputting historical behavior feature data and historical delivery data corresponding to each historical order into the first prediction model, and obtaining a predicted address modification rate corresponding to each historical order output by the first prediction model;
[0032] When the proportion of historical orders with a predicted address modification rate greater than a preset modification rate is greater than a second preset ratio threshold, the parameters of the first prediction model are adjusted, and the step of inputting the historical behavior characteristic data and the historical delivery data corresponding to each historical order into the first prediction model to obtain the predicted address modification rate corresponding to each historical order output by the first prediction model is executed again, until the proportion of historical orders with a predicted address modification rate greater than the preset modification rate is less than or equal to the second preset ratio threshold.
[0033] In a possible implementation, the method further includes:
[0034] Inputting the historical behavior feature data and historical delivery data corresponding to each historical order into the second prediction model, and obtaining the predicted address fulfillment rate corresponding to each historical order output by the second prediction model;
[0035] When the proportion of historical orders whose predicted address fulfillment rate is greater than the preset fulfillment rate is less than the third preset ratio threshold, adjust the parameters of the second prediction model, and again input the historical behavior characteristic data and historical delivery data corresponding to each historical order into the second prediction model to obtain the predicted address fulfillment rate corresponding to each historical order output by the second prediction model, until the proportion of historical orders whose predicted address fulfillment rate is greater than the preset fulfillment rate is greater than or equal to the third preset ratio threshold.
[0036] In a second aspect, an embodiment of the present application provides an address recommendation device, including:
[0037] Recall module, used to obtain multiple address records of target users;
[0038] A rough sorting module, used to roughly sort the multiple address records according to the current positioning longitude and latitude of the target user, and filter the multiple address records based on the result of the rough sorting to obtain a first address list corresponding to at least one filtered first address record;
[0039] A fine sorting module is used to obtain the behavior characteristic data and current delivery data of the target user, and input the behavior characteristic data, current delivery data and the first address list into the scoring model, and fine sort the first address list according to the output result of the scoring model to obtain a second address list; the scoring model is trained by historical order data corresponding to at least one historical order submitted by the target user, and the historical order data includes: historical behavior characteristic data, historical delivery data and historical address records of the target user;
[0040] The sending module is used to send the address information corresponding to the second address list to the terminal corresponding to the target user.
[0041] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a processor and a memory; wherein the memory stores a computer program, and the computer program is suitable for being loaded by the processor and executing the method steps provided in the first aspect of the embodiment of the present application.
[0042] In a fourth aspect, an embodiment of the present application provides a computer storage medium, wherein the computer storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing the method steps provided in the first aspect of the embodiment of the present application.
[0043] The above-mentioned address recommendation method, device, electronic device and computer storage medium realize accurate address recommendation. The method obtains multiple address records of the target user; roughly sorts the multiple address records according to the current positioning longitude and latitude of the target user, and screens the multiple address records based on the result of the rough sorting to obtain a first address list corresponding to at least one first address record screened out; obtains the behavior feature data and current delivery data of the target user, and inputs the behavior feature data, current delivery data and the first address list into the scoring model, and precisely sorts the first address list according to the output result of the scoring model to obtain a second address list; the scoring model is trained by historical order data corresponding to at least one historical order submitted by the target user, and the historical order data includes: the historical behavior feature data, historical delivery data and historical address records of the target user; the address information corresponding to the second address list is sent to the terminal corresponding to the target user. In the first aspect, by combining the current positioning longitude and latitude of the target user, the multiple address records are roughly sorted and screened, so that the recommended address is more in line with the actual needs of the user at the moment, and the practicality of the address recommendation is improved. Secondly, by integrating the output results of the scoring model with the target user's behavioral feature data and the current delivery data, the selected address list is sorted so that the sorted address list is more in line with the user's actual needs and expectations, improving the accuracy of address recommendations. Thirdly, the scoring model is trained based on the target user's historical order data, so that the output results of the scoring model can adapt to the user's personalized needs, effectively ensuring the effectiveness of address recommendations. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments 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 creative work.
[0045] Figure 1 A schematic diagram of an order confirmation page provided for an exemplary embodiment of the present application;
[0046] Figure 2 A schematic diagram of a delivery address selection page provided for an exemplary embodiment of the present application;
[0047] Figure 3 A schematic diagram of an application environment of an address recommendation method provided by an exemplary embodiment of the present application;
[0048] Figure 4 A flowchart of an address recommendation method provided by an exemplary embodiment of the present application;
[0049] Figure 5 A flowchart of another address recommendation method provided by an exemplary embodiment of the present application;
[0050] Figure 6 A schematic diagram of the architecture of an address recommendation method provided by an exemplary embodiment of the present application;
[0051] Figure 7 A flowchart of another address recommendation method provided by an exemplary embodiment of the present application;
[0052] Figure 8 A schematic diagram of the structure of an address recommendation device provided by an exemplary embodiment of the present application;
[0053] Fig. 9 A schematic structural diagram of an electronic device provided as an exemplary embodiment of the present application. DETAILED DESCRIPTION
[0054] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0055] In the description of the present application, it should be understood that the terms "first", "second", etc. are used for descriptive purposes only and should not be understood as indicating or implying relative importance. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances. In addition, in the description of the present application, unless otherwise specified, "multiple" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects associated before and after are in an "or" relationship.
[0056] See also Figure 1 and Figure 2 First, an application scenario applicable to the embodiment of the present application is introduced:
[0057] After a user selects a product using a food delivery app (Application) on a mobile phone or other user terminal, the food delivery app enters a window such as Figure 1 The order confirmation page shown in the figure shows the delivery address information 10 automatically selected by the system based on the current available address. The delivery address information 10 includes the delivery address label, detailed delivery address, consignee and contact number. The user can directly use the delivery address information 10 to submit the order, or click the arrow 11 on the right side of the delivery address information 10 to select other delivery address information to submit the order. After the user clicks the arrow 11, the food delivery APP enters the following Figure 2 The page shown in the figure, in which a floating pop-up window 20 for selecting a delivery address appears by sliding up from the bottom of the screen, covering part of the information displayed on the confirmation order page. In this floating pop-up window 20 for selecting a delivery address, the delivery address information 21~27 recommended by the system are listed in descending order according to the priority order predicted by the system. On the left side of each delivery address information, there is a selection icon corresponding to each delivery address information, and the user can select the desired delivery address information by clicking the selection icon. On the right side of each delivery address information, there is an edit icon corresponding to each delivery address information, and the user can change the corresponding delivery address information by clicking the edit icon, such as changing the delivery address label, detailed delivery address, consignee and contact number in the delivery address information. In addition, there is also a new delivery address control 28 in the floating pop-up window 20 for selecting the delivery address, and the user can add a new delivery address by clicking the new delivery address control 28. After the user selects the desired delivery address information, the floating pop-up window 20 for selecting the delivery address disappears automatically, and the takeaway APP updates the confirmation order page according to the delivery address information selected by the user, and the user can continue to complete the subsequent ordering steps based on the updated confirmation order page.
[0058] The address recommendation method provided in the embodiment of the present application can be applied to Figure 3In the application environment shown. Among them, the user terminal 31 communicates with the takeaway service system 32 through the network. The online storage system can use a key-value method to store the data that the takeaway service system 32 needs to process. The online storage system can be integrated on the takeaway service system 32, or it can be placed on the cloud or other network servers. The takeaway service system 32 obtains multiple address records of the target user from the online storage system; roughly sorts the multiple address records according to the current positioning longitude and latitude of the target user, and screens the multiple address records based on the result of the rough sorting to obtain a first address list corresponding to at least one first address record screened out; obtains the behavior feature data and current delivery data of the target user, and inputs the behavior feature data, current delivery data and the first address list into the scoring model, and finely sorts the first address list according to the output result of the scoring model to obtain a second address list; the scoring model is trained by the historical order data corresponding to at least one historical order submitted by the target user, and the historical order data includes: the historical behavior feature data, historical delivery data and historical address records of the target user; the address information corresponding to the second address list is sent to the terminal corresponding to the target user.
[0059] The user terminal 31 may be, but is not limited to, various smart phones, tablet computers, IoT devices and portable wearable devices. The IoT devices may be smart speakers, smart car-mounted devices, etc. The portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. The food delivery service system 32 may be implemented using an independent server or a server cluster consisting of multiple servers.
[0060] In one embodiment, Figure 4 As shown, an address recommendation method is provided, which is applied to Figure 1 Taking the takeaway service system 32 in the example as an example, the following steps are included:
[0061] S401: Acquire multiple address records of a target user.
[0062] Among them, multiple address records include address records corresponding to the delivery addresses created by the target user in the food delivery APP and address records corresponding to the target user’s historical delivery addresses; each address record includes the delivery address label (such as company, home, smart cabinet, etc.), detailed delivery address (such as community name, house number, etc.), consignee, contact number, delivery address latitude and longitude, and whether the delivery address is valid and other delivery address information.
[0063] Optionally, the food delivery service system 32 imports the address records corresponding to the delivery addresses created by the user in the food delivery APP and the address records corresponding to the historical delivery addresses into an online storage system (such as Redis) in the form of full or incremental data, and stores them with the user's user identifier (such as user ID) as a key value. Each address record is an independent entry, containing complete delivery address information such as the delivery address label, detailed delivery address, consignee, contact number, delivery address latitude and longitude, and whether the delivery address is valid. When the target user enters such Figure 1 After the order confirmation page is displayed, the food delivery service system 32 obtains the full amount of delivery address information of the user from the online storage system according to the user ID of the target user, that is, multiple address records corresponding to all delivery addresses saved by the user in the food delivery APP.
[0064] S402: Roughly sorting the plurality of address records according to the current positioning longitude and latitude of the target user, and filtering the plurality of address records based on the result of the rough sorting to obtain a first address list corresponding to at least one filtered first address record.
[0065] Among them, each address record includes the delivery address information such as the longitude and latitude of the delivery address and whether the delivery address is valid.
[0066] Optionally, the current positioning longitude and latitude of the target user is first obtained through satellite positioning, network positioning and other technologies, and the longitude and latitude distance between the current positioning longitude and latitude and longitude and the longitude and latitude of the delivery address corresponding to each address record is calculated. Based on the calculated longitude and latitude distance, the multiple address records of the target user are roughly sorted in order from near to far, and multiple address records with longitude and latitude distances less than a preset longitude and latitude distance (such as 3 kilometers) are retained, and the validity of the addresses is further determined for the multiple retained address records.
[0067] Specifically, determining the validity of an address includes determining whether the delivery address corresponding to the address record is within the delivery range of the merchant selected by the user, and determining whether the delivery address corresponding to the address record is open to the public. By determining the validity of the address on the retained address records, a first address list corresponding to at least one first address record is screened out.
[0068] S403: Obtain the behavior characteristic data and current delivery data of the target user, and input the behavior characteristic data, current delivery data and the first address list into the scoring model, and finely sort the first address list according to the output result of the scoring model to obtain the second address list.
[0069] Optionally, the target user's behavioral characteristic data includes the target user's transaction behavior data, the target user's browsing behavior data, and the target user's purchase preference data. Among them, the target user's transaction behavior data specifically includes the target user's order time (such as weekday evenings, weekend mornings, etc.), the target user's order frequency (such as placing orders every day, etc.), and the target user's address switching frequency when selecting a delivery address (such as frequent switching when selecting a delivery address, etc.); the target user's browsing behavior data specifically includes the target user's browsing times on the product page, the target user's stay time on the product page, and the target user's click times on the product; the target user's purchase preference data specifically includes the product information of the target user's frequently purchased products and the product information of the target user's additional purchases. The current delivery data includes the current weather data and the current road condition data.
[0070] Specifically, the food delivery APP imports the user's behavioral characteristic data into an online storage system (such as Redis) in the form of full or incremental data, and stores it with the user's user identification (such as user ID) key value. When the target user enters the order confirmation page, the food delivery service system 32 obtains the target user's behavioral characteristic data from the online storage system according to the target user's user identification, and obtains the current delivery data through the real-time data interface. Then, the target user's behavioral characteristic data, current delivery data, and the first address list are input into the scoring model, and the first address list is precisely sorted in real time according to the results output by the scoring model.
[0071] It is worth noting that the scoring model is trained by historical order data corresponding to at least one historical order submitted by the target user, and the historical order data includes: historical behavior feature data of the target user, historical delivery data and historical address records. Among them, the historical behavior feature data of the target user includes but is not limited to the historical transaction behavior data of the target user, the historical browsing behavior data of the target user and the historical purchase preference data of the target user. The historical transaction behavior data of the target user includes but is not limited to the historical order time of the target user, the historical order frequency of the target user and the address switching frequency when the target user selects the historical delivery address; the historical browsing behavior data of the target user specifically includes but is not limited to the historical browsing times of the target user on the product page, the historical stay time of the target user on the product page, and the historical click times of the target user on the product; the historical purchase preference data of the target user includes but is not limited to the product information of the target user's historical frequently purchased products and the product information of the target user's historical added purchase products. The historical delivery data includes but is not limited to the historical weather data and the historical road condition data. The historical address record refers to the address record corresponding to the historical delivery address used by the user.
[0072] S404: Send the address information corresponding to the second address list to the terminal corresponding to the target user.
[0073] Among them, the address information corresponding to the second address list includes delivery address information such as delivery address label (such as company, home, smart cabinet, etc.), detailed delivery address (such as community name, house number, etc.), consignee, contact number, etc.
[0074] Optionally, after obtaining the second address list, the takeaway service system 32 first combines the address records stored in the online storage system to obtain the address information corresponding to the second address list, such as the delivery address label, detailed delivery address, consignee, contact number, etc. Then, the address information corresponding to the second address list is packaged into a data packet in JSON format or other data formats. Finally, according to the user identifier of the target user (such as user ID, device ID, etc.), the packaged address information data packet is sent to the terminal corresponding to the target user (such as a smart phone, tablet computer, etc.) through the network interface. After receiving the address information data packet sent by the server, the terminal corresponding to the target user parses it, and uses frameworks such as React, Vue, Angular, etc. for front-end rendering, and updates the delivery address information displayed on the confirmation order page.
[0075] The above-mentioned address recommendation method, by combining the current location latitude and longitude of the target user, roughly sorts and filters multiple address records, so that the recommended addresses are more in line with the user's current actual needs, thereby improving the practicality of address recommendation; by integrating the output results of the scoring model of the target user's behavioral feature data and the current delivery data, the filtered address list is finely sorted, so that the finely sorted address list is more in line with the user's actual needs and expectations, thereby improving the accuracy of address recommendation; the scoring model is trained based on the historical order data of the target user, so that the output results of the scoring model can adapt to the user's personalized needs, effectively ensuring the effectiveness of address recommendation. The above-mentioned address recommendation method can achieve accurate address recommendation.
[0076] In one embodiment, Figure 5 As shown, another address recommendation method is provided, which is applied to Figure 1 Taking the takeaway service system 32 in the example as an example, the following steps are included:
[0077] S501: Acquire multiple address records of a target user; each address record in the multiple address records corresponds to an address index.
[0078] Among them, each address record in the multiple address records includes address latitude and longitude information and address validity information; the address latitude and longitude information is used to represent the address latitude and longitude corresponding to each address record, and the address validity information is used to represent whether the address corresponding to each address record is valid.
[0079] It is understandable that users generally save multiple delivery addresses (such as home addresses, company addresses, or temporarily created addresses) on the food delivery APP by filling in or selecting. If the delivery address filled in or selected by the user is a known point of interest with a unique identifier in the POI (Point of Interest) database, the food delivery service system 32 can directly extract the address longitude and latitude information (such as longitude and latitude coordinates) and address validity information (such as whether the address actually exists or whether the address is open to the public) corresponding to the delivery address from the POI database. If the delivery address filled in or selected by the user is not in the POI database, the food delivery service system 32 will obtain the longitude and latitude information and address validity information of the address based on the integrated map API (Application Programming Interface) and / or third-party address verification service.
[0080] Optionally, after the food delivery service system 32 obtains the address latitude and longitude information, address validity information and other address information (such as the delivery address label, detailed delivery address, consignee, contact number, etc.) corresponding to the delivery address saved by the user, it will import the address record containing the above address information as a separate entry into the online storage system and store it in correspondence with the user identifier (such as the user ID). The online storage system will assign a unique identifier, i.e., an address index (such as an address ID), to each address record to help the food delivery service system 32 identify and distinguish the address records corresponding to the user's different delivery addresses, thereby improving the efficiency of data retrieval and subsequent processing. When the target user enters a page such as Figure 1 After the order confirmation page is displayed, the food delivery service system 32 obtains the full delivery address information of the target user from the online storage system, that is, multiple address records corresponding to all delivery addresses saved by the target user in the food delivery APP. Each address record includes address latitude and longitude information used to represent the longitude and latitude of the address and address validity information used to represent whether the address is valid, and each address record includes an address index uniquely corresponding to it.
[0081] S502: Calculate the longitude and latitude distance between the current positioning longitude and latitude of the target user and the address longitude and latitude contained in each address record.
[0082] Among them, the current positioning longitude and latitude refer to the positioning longitude and latitude returned by the built-in positioning module of the user terminal (such as a mobile phone) of the target user. The current positioning longitude and latitude can be but are not limited to being obtained by relying on satellite positioning, network positioning and other technologies; the address longitude and latitude contained in each address record refers to the longitude and latitude corresponding to the delivery address saved by the target user returned by the POI database and / or map API and / or third-party address verification service; the longitude and latitude distance can be but are not limited to the actual longitude and latitude distance that takes into account the curvature of the earth.
[0083] Optionally, the food delivery APP obtains the current positioning longitude and latitude returned by the user terminal of the target user by initiating a positioning service, and sends the current positioning longitude and latitude to the food delivery service system 32. The food delivery service system 32 uses the Haversine formula or the cosine formula to calculate the actual longitude and latitude distance between the current positioning longitude and latitude and the address longitude and latitude in each address record.
[0084] S503: Filter out address records whose longitude and latitude distance is less than or equal to the preset longitude and latitude distance and whose addresses are valid.
[0085] Among them, the preset longitude and latitude distance is a pre-set longitude and latitude distance threshold, which depends on the delivery range of the food delivery platform. For example, it can be set to 3 kilometers, and it can also be adjusted according to actual conditions, such as adjusting it to 2 kilometers, 4 kilometers, 5 kilometers, etc.
[0086] Optionally, the food delivery service system 32 retains address records whose longitude and latitude distance is less than or equal to a preset longitude and latitude distance (such as 3 kilometers), and screens out address records whose addresses are real and valid (such as the address actually exists and is open to the public) based on the address validity information contained in the retained address records.
[0087] S504: Obtain a first address list corresponding to at least one filtered first address record; the first address list is constructed by address indexes corresponding to each first address record.
[0088] Optionally, after filtering out the first address records whose longitude and latitude distances are less than or equal to the preset longitude and latitude distances and whose addresses are valid, the food delivery service system 32 does not directly use at least one of the filtered out first address records to construct a first address list, but obtains the address index corresponding to each of the at least one first address record and constructs the first address list through at least one address index.
[0089] It is understandable that, since each address record actually corresponds to a unique address index, the address index can effectively represent the complex address information in each first address record. The first address list is constructed using the address index corresponding to each first address record, so that subsequent processing and analysis can be focused on a lighter index, reducing unnecessary data handling and reducing the complexity of data processing.
[0090] S505: Obtain the behavior characteristic data and current delivery data of the target user, and input the behavior characteristic data, current delivery data and the first address list into the scoring model to obtain the address recommendation rate of each first address record corresponding to the first address list output by the scoring model.
[0091] Among them, behavioral feature data include but are not limited to transaction behavior data, browsing behavior data and purchase preference data. Transaction behavior data include but are not limited to order time, order frequency and address switching frequency, browsing behavior data include but are not limited to the number of views on the product page, the dwell time on the product page and the number of clicks on the product. Purchase preference data include but are not limited to the product information of frequently purchased products and the product information of added products. Current delivery data include but are not limited to current weather data and current road condition data. The scoring model is a supervised learning model that uses the address recommendation rate as a label.
[0092] Optionally, see Figure 6 The food delivery service system 32 obtains the behavioral characteristic data and current delivery data of the target user, and inputs the behavioral characteristic data, current delivery data and the first address list constructed by the address index corresponding to each first address record into a scoring model that uses the address recommendation rate as a label, and obtains the address recommendation rate corresponding to each address index in the first address list, that is, the address recommendation rate of each first address record corresponding to the first address list.
[0093] In one embodiment, the above-mentioned address recommendation method also includes: inputting the historical behavior characteristic data, historical delivery data and historical address records corresponding to each historical order into the scoring model to obtain the predicted address recommendation rate corresponding to each historical address record output by the scoring model; when the proportion of historical orders whose predicted address recommendation rate is greater than the preset recommendation rate is less than the first preset ratio threshold, adjusting the parameters of the scoring model, and inputting the historical behavior characteristic data, historical delivery data and historical address records corresponding to each historical order into the scoring model again to obtain the predicted address recommendation rate corresponding to each historical address record output by the scoring model, until the proportion of historical orders whose predicted address recommendation rate is greater than the preset recommendation rate is greater than or equal to the first preset ratio threshold.
[0094] Among them, historical behavior feature data includes but is not limited to historical transaction behavior data, historical browsing behavior data and historical purchase preference data. Historical transaction behavior data includes but is not limited to historical order time, historical order frequency and address switching frequency. Historical browsing behavior data includes but is not limited to the historical number of views on the product page, the historical stay time on the product page, and the historical number of clicks on the product. Historical purchase preference data includes but is not limited to the product information of historical frequently purchased products and historical product information of added products. Historical delivery data includes but is not limited to historical weather data and historical road condition data. Historical address records refer to the address records corresponding to the historical delivery addresses used by users.
[0095] Optionally, the scoring model can be, but is not limited to, a supervised learning model based on a multilayer perceptron (MLP) algorithm or a gradient boosting decision tree (GBDT) algorithm or a deep neural network (DNN) algorithm, which is trained using the address recommendation rate as a label. Since the historical delivery addresses corresponding to each historical address record are the delivery addresses actually selected by the user, the preset recommendation rate can be set to a higher value (such as 98%). In the process of training the scoring model, the parameters of the scoring model are continuously adjusted so that the proportion of historical orders with a predicted address recommendation rate greater than the preset recommendation rate is greater than or equal to a first preset ratio threshold (such as 90%), so that the trained scoring model can output an address recommendation rate close to the user's selection.
[0096] S506: Sort the first address list in descending order based on the address recommendation rate to obtain a second address list.
[0097] Optionally, the food delivery service system 32 sorts the first address list in descending order according to the address recommendation rate corresponding to each address index in the first address list to obtain a second address list.
[0098] It can be understood that by constructing a first address list based on address indexes and generating a second address list by sorting them according to the behavioral characteristics of different users and the current delivery conditions, the arrangement of the address list can be effectively optimized and the user's interactive experience can be improved.
[0099] S507: Complete the address information according to the address index in the second address list to obtain the address information corresponding to the second address list.
[0100] Optionally, since the second address list is obtained by sorting the first address list constructed by the address indexes corresponding to the first address records in descending order based on the address recommendation rate, the takeaway service system 32 can complete the address information according to the address indexes in the second address list, such as completing the delivery address label, detailed delivery address, consignee, contact number, address longitude and latitude information, address validity information, etc.
[0101] It can be understood that the address information corresponding to the second address list obtained after completion is the delivery address information recommended to the user. The address information is completed by the address index, which effectively ensures the integrity of the delivery address information recommended to the user.
[0102] S508: Send the address information corresponding to the second address list to the terminal corresponding to the target user.
[0103] Specifically, S508 is consistent with S404 and will not be described in detail here.
[0104] In this embodiment, on the one hand, the addresses are screened according to the user's current location and address validity, ensuring that the recommended addresses are authentic, valid and deliverable, thereby improving the practicality of address recommendations; on the second hand, the first list of screened addresses is sorted in descending order using a scoring model in combination with the target user's behavioral feature data and current delivery data, thereby making the recommended addresses more in line with the user's actual needs and current delivery conditions, thereby improving the accuracy of address recommendations; on the third hand, the address index corresponding to the first address record is sorted in descending order, and the address information of the second address list obtained after descending sorting is completed, thereby simplifying the data processing flow while ensuring the integrity of the delivery address information recommended to the user, thereby improving the efficiency of address recommendations.
[0105] In one embodiment, Figure 7 As shown, after obtaining the target user's behavior feature data and current delivery data, the above-mentioned address recommendation method further includes the following steps:
[0106] S701: Input the behavioral feature data and the current delivery data into the prediction model to obtain the address modification rate and address fulfillment rate corresponding to the current order submitted by the target user; the prediction model includes a first prediction model for predicting the order address modification rate and a second prediction model for predicting the order address fulfillment rate; the first prediction model and the second prediction model are trained by historical order data.
[0107] Optionally, the takeaway service system 32 extracts the target user's behavioral feature data (such as order time, number of views of the product page, product information of frequently purchased products, etc.) from the data storage system according to the user identification (such as user ID), and obtains the current delivery data (such as current weather data, current road condition data, etc.) through the real-time data interface. Then, the behavioral feature data and the current delivery data are respectively input into the first prediction model and the second prediction model, and the first prediction model is used to predict the address modification rate corresponding to the current order submitted by the target user, and the second prediction model is used to predict the address fulfillment rate corresponding to the current order submitted by the target user. The above-mentioned first prediction model and the second prediction model are both trained by historical order data, and the historical order data includes but is not limited to the historical behavioral feature data and historical delivery data corresponding to each historical order.
[0108] In one embodiment, the above-mentioned address recommendation method also includes: inputting the historical behavior characteristic data and historical delivery data corresponding to each historical order into the first prediction model to obtain the predicted address modification rate corresponding to each historical order output by the first prediction model; when the proportion of historical orders whose predicted address modification rate is greater than the preset modification rate is greater than the second preset ratio threshold, adjusting the parameters of the first prediction model, and inputting the historical behavior characteristic data and historical delivery data corresponding to each historical order into the first prediction model again to obtain the predicted address modification rate corresponding to each historical order output by the first prediction model, until the proportion of historical orders whose predicted address modification rate is greater than the preset modification rate is less than or equal to the second preset ratio threshold.
[0109] The historical behavior characteristic data corresponding to each historical order includes, but is not limited to, historical order time, historical order frequency, historical browsing times on the product page, product information of historical frequently purchased products, etc. The historical delivery data includes, but is not limited to, historical weather data and historical road condition data.
[0110] Optionally, the first prediction model may be, but is not limited to, a supervised learning model based on a multilayer perceptron (MLP) algorithm or a gradient boosting decision tree (GBDT) algorithm or a deep neural network (DNN) algorithm, which is trained using the address modification rate as a label. The address modification rate refers to the proportion of the number of historical orders with modified addresses to the total number of historical orders, which is used to characterize the frequency of address modification by the target user. The higher the address modification rate of the target user, the greater the possibility that the target user will modify the delivery address after submitting the order.
[0111] Specifically, when training the first prediction model, firstly, the preset modification rate is calculated according to the address modification situations corresponding to the multiple historical orders participating in the training. The calculation formula of the preset modification rate is as follows:
[0112] (1)
[0113] After calculating the preset modification rates corresponding to multiple historical orders participating in the training, the historical behavior feature data and historical delivery data corresponding to each historical order are input into the first prediction model, and the parameters of the first prediction model are continuously adjusted so that the proportion of historical orders with a predicted address modification rate greater than the preset modification rate is less than or equal to a second preset ratio threshold (such as 10%), so that the trained first prediction model can output a more accurate predicted address modification rate.
[0114] In this embodiment, the parameters of the first prediction model are optimized through continuous iteration until the predicted address modification rate output by the first prediction model is more consistent with the actual address modification of the user, thereby effectively improving the accuracy of the model prediction.
[0115] In one embodiment, the above-mentioned address recommendation method also includes: inputting the historical behavior characteristic data and historical delivery data corresponding to each historical order into the second prediction model to obtain the predicted address fulfillment rate corresponding to each historical order output by the second prediction model; when the proportion of historical orders whose predicted address fulfillment rate is greater than the preset fulfillment rate is less than a third preset ratio threshold, adjusting the parameters of the second prediction model, and inputting the historical behavior characteristic data and historical delivery data corresponding to each historical order into the second prediction model again to obtain the predicted address fulfillment rate corresponding to each historical order output by the second prediction model, until the proportion of historical orders whose predicted address fulfillment rate is greater than the preset fulfillment rate is greater than or equal to the third preset ratio threshold.
[0116] Optionally, the second prediction model may be, but is not limited to, a supervised learning model based on a multilayer perceptron (MLP) algorithm or a gradient boosting decision tree (GBDT) algorithm or a deep neural network (DNN) algorithm, which is trained using the address fulfillment rate indicator as a label. The address fulfillment rate refers to the ratio of the number of historical orders that have been successfully fulfilled to the total number of historical orders, which is used to characterize the probability that the order submitted by the target user can be successfully delivered. The higher the address fulfillment rate of the target user, the more likely it is that the order submitted by the target user can be successfully delivered.
[0117] Specifically, when training the second prediction model, firstly, the preset fulfillment rate is calculated according to the address fulfillment status corresponding to the multiple historical orders participating in the training. The calculation formula of the preset fulfillment rate is as follows:
[0118] (2)
[0119] After calculating the preset fulfillment rates corresponding to multiple historical orders participating in the training, the historical behavioral feature data and historical delivery data corresponding to each historical order are input into the second prediction model, and the parameters of the second prediction model are continuously adjusted so that the proportion of historical orders with a predicted address fulfillment rate greater than the preset fulfillment rate is greater than or equal to a third preset ratio threshold (such as 90%), so that the trained second prediction model can output a more accurate predicted address fulfillment rate.
[0120] In this embodiment, the parameters of the second prediction model are optimized through continuous iteration until the predicted address fulfillment rate output by the second prediction model is more consistent with the user's actual address fulfillment situation, thereby effectively improving the accuracy of the model prediction.
[0121] S702: When the address modification rate corresponding to the current order is greater than the preset modification rate, or the address fulfillment rate corresponding to the current order is less than the preset fulfillment rate, an address input prompt message is sent to a terminal corresponding to the target user.
[0122] The address input prompt information is used to prompt the user to manually fill in or select the delivery address.
[0123] Optionally, if the address modification rate corresponding to the current order is greater than the preset modification rate, it means that the user is very likely to modify the delivery address after submitting the order. If the address fulfillment rate corresponding to the current order is less than the preset fulfillment rate, it means that the order submitted by the user is very likely to fail to be delivered. Therefore, in the above two cases, the food delivery service system 32 will send the address input prompt information to the terminal corresponding to the target user to prompt the user to manually fill in or select the delivery address.
[0124] S703: When the address modification rate corresponding to the current order is less than or equal to the preset modification rate, or the address fulfillment rate corresponding to the current order is greater than or equal to the preset fulfillment rate, an address confirmation prompt message is sent to the terminal corresponding to the target user.
[0125] The address input confirmation information is used to prompt the user to confirm the selected delivery address.
[0126] Optionally, if the address modification rate corresponding to the current order is less than or equal to the preset modification rate, it means that the user is very likely not to modify the delivery address after submitting the order. If the address fulfillment rate corresponding to the current order is less than the preset fulfillment rate, it means that the order submitted by the user has a high probability of being successfully delivered. Therefore, in the above two cases, the food delivery service system 32 will not send the address input prompt information to the terminal corresponding to the target user, but will send the address confirmation prompt information to the terminal corresponding to the target user to prompt the user to confirm the selected delivery address.
[0127] In this embodiment, on the one hand, by inputting the historical order data of the target user into the prediction model, and dynamically adjusting the model parameters according to the deviation between the prediction results (predicted address modification rate and predicted address fulfillment rate) and the actual performance (preset modification rate and preset fulfillment rate), the model is continuously optimized and the prediction accuracy of the model is improved. On the other hand, the system can actively send prompt information to the user based on the output of the prediction model. If it is predicted that the delivery address of the current order may need to be modified or the fulfillment risk is high, the user will be prompted to manually select or enter the delivery address. If it is predicted that the delivery address of the current order may not need to be modified or the fulfillment risk is low, the user will be prompted to confirm the address information, thereby reducing subsequent delivery problems and improving the user experience.
[0128] Based on the inventive concept of the above-mentioned address recommendation method, Figure 8 As shown, the embodiment of the present application further provides an address recommendation device 800 for implementing the above-mentioned address recommendation method. The address recommendation device 800 includes:
[0129] The recall module 801 is used to obtain multiple address records of the target user;
[0130] A rough sorting module 802 is used to roughly sort the multiple address records according to the current positioning longitude and latitude of the target user, and filter the multiple address records based on the result of the rough sorting to obtain a first address list corresponding to at least one filtered first address record;
[0131] The fine sorting module 803 is used to obtain the behavior characteristic data and current delivery data of the target user, and input the behavior characteristic data, current delivery data and the first address list into the scoring model, and fine sort the first address list according to the output result of the scoring model to obtain the second address list; the scoring model is trained by the historical order data corresponding to at least one historical order submitted by the target user, and the historical order data includes: the historical behavior characteristic data, historical delivery data and historical address records of the target user;
[0132] The sending module 804 is used to send the address information corresponding to the second address list to the terminal corresponding to the target user.
[0133] In one embodiment, each address record in the multiple address records includes address longitude and latitude information and address validity information; the address longitude and latitude information is used to characterize the address longitude and latitude corresponding to each address record, and the address validity information is used to characterize whether the address corresponding to each address record is valid; the above-mentioned coarse sorting module 802 is also used to calculate the longitude and latitude distance between the current positioning longitude and latitude of the target user and the address longitude and latitude contained in each address record; and filter out address records whose longitude and latitude distance is less than or equal to the preset longitude and latitude distance and whose addresses are valid.
[0134] In one embodiment, each address record in the plurality of address records corresponds to an address index; the first address list is constructed by the address index corresponding to each first address record; the above-mentioned fine sorting module 803 is also used to complete the address information according to the address index in the second address list to obtain the address information corresponding to the second address list.
[0135] In one embodiment, the above-mentioned fine sorting module 803 is also used to input the behavioral characteristic data, the current delivery data and the first address list into the scoring model to obtain the address recommendation rate of each first address record corresponding to the first address list output by the scoring model; sort the first address list in descending order based on the address recommendation rate to obtain the second address list.
[0136] In one embodiment, the above-mentioned fine sorting module 803 is also used to input the historical behavior characteristic data, historical delivery data and historical address records corresponding to each historical order into the scoring model, and obtain the predicted address recommendation rate corresponding to each historical address record output by the scoring model; when the proportion of historical orders whose predicted address recommendation rate is greater than the preset recommendation rate is less than the first preset ratio threshold, adjust the parameters of the scoring model, and input the historical behavior characteristic data, historical delivery data and historical address records corresponding to each historical order into the scoring model again to obtain the predicted address recommendation rate corresponding to each historical address record output by the scoring model, until the proportion of historical orders whose predicted address recommendation rate is greater than the preset recommendation rate is greater than or equal to the first preset ratio threshold.
[0137] In one embodiment, the address recommendation device 800 further includes a prediction module for inputting behavioral feature data and current delivery data into a prediction model to obtain an address modification rate and an address fulfillment rate corresponding to a current order submitted by a target user; the prediction model includes a first prediction model for predicting an order address modification rate and a second prediction model for predicting an order address fulfillment rate; the first prediction model and the second prediction model are trained by historical order data; when the address modification rate corresponding to the current order is greater than a preset modification rate, or the address fulfillment rate corresponding to the current order is less than a preset fulfillment rate, an address input prompt message is sent to a terminal corresponding to the target user; when the address modification rate corresponding to the current order is less than or equal to a preset modification rate, or the address fulfillment rate corresponding to the current order is greater than or equal to a preset fulfillment rate, an address confirmation prompt message is sent to a terminal corresponding to the target user.
[0138] In one embodiment, the above-mentioned prediction module is also used to input the historical behavior characteristic data and historical delivery data corresponding to each historical order into the first prediction model to obtain the predicted address modification rate corresponding to each historical order output by the first prediction model; when the proportion of historical orders whose predicted address modification rate is greater than the preset modification rate is greater than the second preset ratio threshold, adjust the parameters of the first prediction model, and input the historical behavior characteristic data and historical delivery data corresponding to each historical order into the first prediction model again to obtain the predicted address modification rate corresponding to each historical order output by the first prediction model, until the proportion of historical orders whose predicted address modification rate is greater than the preset modification rate is less than or equal to the second preset ratio threshold.
[0139] In one embodiment, the above-mentioned prediction module is also used to input the historical behavior characteristic data and historical delivery data corresponding to each historical order into the second prediction model to obtain the predicted address fulfillment rate corresponding to each historical order output by the second prediction model; when the proportion of historical orders whose predicted address fulfillment rate is greater than the preset fulfillment rate is less than a third preset ratio threshold, adjust the parameters of the second prediction model, and input the historical behavior characteristic data and historical delivery data corresponding to each historical order into the second prediction model again to obtain the predicted address fulfillment rate corresponding to each historical order output by the second prediction model, until the proportion of historical orders whose predicted address fulfillment rate is greater than the preset fulfillment rate is greater than or equal to the third preset ratio threshold.
[0140] Each module in the above-mentioned address recommendation device 800 can be implemented in whole or in part by software, hardware or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each module.
[0141] The embodiment of the present application also provides an electronic device, which may be a server, and its internal structure diagram may be as follows: Fig. 9As shown. The electronic device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the electronic device is used to store multiple address records and historical order data of the target user. The input / output interface of the electronic device is used to exchange information between the processor and an external device. The communication interface of the electronic device is used to communicate with an external terminal through a network connection. The processor of the electronic device executes a computer program to implement an address recommendation method.
[0142] Those skilled in the art will understand that Fig. 9 The structure shown in the figure is merely a block diagram of a partial structure related to the scheme of the present application, and does not constitute a limitation on the electronic device to which the scheme of the present application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.
[0143] The present application also provides a computer storage medium, which stores instructions, and when the instructions are executed on a computer or a processor, the computer or the processor executes one or more steps in the above embodiment. If the components of the above electronic device are implemented in the form of software functional units and sold or used as independent products, they can be stored in the above computer readable storage medium.
[0144] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer storage medium or transmitted through the computer storage medium. The computer instructions may be transmitted from a website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a digital versatile disc (DVD)), or a semiconductor medium (eg, a solid state drive (SSD)).
[0145] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk and other media that can store program codes. In the absence of conflict, the technical features in this embodiment and the implementation scheme can be combined arbitrarily.
[0146] The embodiments described above are merely preferred embodiments of the present application and are not intended to limit the scope of the present application. Without departing from the design spirit of the present application, various modifications and improvements made to the technical solutions of the present application by ordinary technicians in this field should fall within the protection scope determined by the claims.
[0147] The above describes a specific embodiment of the present application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
Claims
1. An address recommendation method, characterized in that: The method comprises: Get multiple address records of the target user; Roughly sorting the multiple address records according to the current positioning longitude and latitude of the target user, and filtering the multiple address records based on the result of the rough sorting to obtain a first address list corresponding to at least one filtered first address record; Obtaining the behavior characteristic data and current delivery data of the target user, and inputting the behavior characteristic data, the current delivery data and the first address list into a scoring model, and performing fine sorting on the first address list according to an output result of the scoring model to obtain a second address list; the scoring model is trained by historical order data corresponding to at least one historical order submitted by the target user, and the historical order data includes: historical behavior characteristic data, historical delivery data and historical address records of the target user; Sending the address information corresponding to the second address list to the terminal corresponding to the target user; After obtaining the behavior characteristic data of the target user and the current delivery data, the method further includes: The behavioral characteristic data and the current delivery data are input into the prediction model to obtain the address modification rate and address fulfillment rate corresponding to the current order submitted by the target user; the prediction model includes a first prediction model for predicting the order address modification rate and a second prediction model for predicting the order address fulfillment rate; the first prediction model and the second prediction model are trained by the historical order data.
2. The method according to claim 1, characterized in that Each of the multiple address records includes address latitude and longitude information and address validity information; the address latitude and longitude information is used to represent the address latitude and longitude corresponding to each address record, and the address validity information is used to represent whether the address corresponding to each address record is valid; The roughly sorting the plurality of address records according to the current positioning longitude and latitude of the target user, and filtering the plurality of address records based on the result of the roughly sorting, comprises: Calculating the longitude and latitude distance between the current positioning longitude and latitude of the target user and the address longitude and latitude contained in each address record; Filter out the address records whose longitude and latitude distance is less than or equal to the preset longitude and latitude distance and whose addresses are valid.
3. The method according to claim 1, characterized in that Each address record in the plurality of address records corresponds to an address index; the first address list is constructed by the address indexes corresponding to each of the first address records; After finely sorting the first address list to obtain the second address list, the method further includes: The address information is completed according to the address index in the second address list to obtain the address information corresponding to the second address list.
4. The method according to claim 1, characterized in that The step of inputting the behavior feature data, the current delivery data, and the first address list into a scoring model, and finely sorting the first address list according to an output result of the scoring model to obtain a second address list includes: Inputting the behavior characteristic data, the current delivery data and the first address list into a scoring model, and obtaining an address recommendation rate of each first address record corresponding to the first address list output by the scoring model; The first address list is sorted in descending order based on the address recommendation rate to obtain a second address list.
5. The method according to claim 1, characterized in that The method further comprises: Inputting the historical behavior feature data, the historical delivery data, and the historical address records corresponding to each of the historical orders into the scoring model, and obtaining a predicted address recommendation rate corresponding to each of the historical address records output by the scoring model; When the proportion of historical orders with the predicted address recommendation rate greater than the preset recommendation rate is less than a first preset ratio threshold, the parameters of the scoring model are adjusted, and the step of inputting the historical behavior characteristic data, the historical delivery data and the historical address records corresponding to each of the historical orders into the scoring model to obtain the predicted address recommendation rate corresponding to each of the historical address records output by the scoring model is performed again, until the proportion of historical orders with the predicted address recommendation rate greater than the preset recommendation rate is greater than or equal to the first preset ratio threshold.
6. The method according to claim 1, characterized in that After obtaining the behavior characteristic data of the target user and the current delivery data, the method further includes: When the address modification rate corresponding to the current order is greater than the preset modification rate, or the address fulfillment rate corresponding to the current order is less than the preset fulfillment rate, sending the address input prompt information to the terminal corresponding to the target user; When the address modification rate corresponding to the current order is less than or equal to the preset modification rate, or the address fulfillment rate corresponding to the current order is greater than or equal to the preset fulfillment rate, an address confirmation prompt message is sent to the terminal corresponding to the target user.
7. The method according to claim 6, characterized in that The method further comprises: Inputting the historical behavior feature data and the historical delivery data corresponding to each of the historical orders into the first prediction model, and obtaining the predicted address modification rate corresponding to each of the historical orders output by the first prediction model; When the proportion of historical orders with the predicted address modification rate greater than the preset modification rate is greater than a second preset ratio threshold, the parameters of the first prediction model are adjusted, and the step of inputting the historical behavior characteristic data and the historical delivery data corresponding to each historical order into the first prediction model to obtain the predicted address modification rate corresponding to each historical order output by the first prediction model is performed again, until the proportion of historical orders with the predicted address modification rate greater than the preset modification rate is less than or equal to the second preset ratio threshold.
8. The method according to claim 6, characterized in that The method further comprises: Inputting the historical behavior feature data and the historical delivery data corresponding to each of the historical orders into the second prediction model, and obtaining the predicted address fulfillment rate corresponding to each of the historical orders output by the second prediction model; When the proportion of historical orders whose predicted address fulfillment rate is greater than the preset fulfillment rate is less than a third preset ratio threshold, the parameters of the second prediction model are adjusted, and the step of inputting the historical behavior characteristic data and the historical delivery data corresponding to each historical order into the second prediction model to obtain the predicted address fulfillment rate corresponding to each historical order output by the second prediction model is performed again, until the proportion of historical orders whose predicted address fulfillment rate is greater than the preset fulfillment rate is greater than or equal to the third preset ratio threshold.
9. An electronic device, characterized in that: include: A processor and a memory; the memory stores a computer program, and the processor implements the method steps of any one of claims 1 to 8 when executing the computer program.
10. A computer storage medium, characterized in that: The computer storage medium stores a plurality of instructions, which are suitable for being loaded by a processor and executing the method steps according to any one of claims 1 to 8.
Citation Information
Patent Citations
Information recommendation method, device and equipment and computer readable storage medium
CN112559896A
Control method, display method and electronic equipment
CN113590893A
Address recommendation method, system and device and storage medium
CN113592401A
Method and device for obtaining recommended form and electronic equipment
CN114385931A
Address matching method and device, equipment and medium
CN116402040A
