Peripheral merchant recommendation method and system in business operation, medium, product

Abstract

The application provides a business peripheral merchant recommendation method and system, a computer readable storage medium, and a computer program product. For each outbound user triggering an outbound behavior, the outbound port to which the outbound behavior belongs is determined according to a user itinerary generated by the outbound user, the peripheral merchants radiating from the outbound port are estimated for conversion value of the peripheral merchants to the user, and the peripheral merchants of the outbound user are pushed by the conversion value of the peripheral merchants to the user. Thus, the peripheral merchants are actively pushed to the outbound user, and the goods of the pushed peripheral merchants are actively exposed. The exposure of the goods of the peripheral merchants is no longer limited to passive triggering of the user. The exposed goods are interesting to the outbound user. With the active exposure, a purchase event is most likely to occur, and the value conversion of the goods is generated.

Description

Technical Field

[0001] The present application relates to the field of computer application technology, and specifically to a method and system for recommending surrounding businesses in commercial operations, a computer-readable storage medium, and a computer program product. Background Art

[0002] There are many businesses distributed around travel stations, such as subways, buses, high-speed railways, airports, ships and other travel stations. Surrounding businesses expose their products to outgoing users, which enables users to make purchases at surrounding businesses, and then the surrounding businesses generate conversions by delivering products to users.

[0003] Although surrounding businesses in commercial operations are within the activity range of users who travel by specific means of transportation, the exposure of relevant information of surrounding businesses to users depends on the exposure of offline physical locations and searches actively initiated by users in terminal devices based on geographic locations. This will make it extremely difficult to expose surrounding businesses to outbound users.

[0004] If the outbound user does not visit nearby merchants, or does not trigger a search for nearby merchants based on the current geographic location through the nearby merchant function of online location service applications or local life applications, then even if the user is interested in the products of the nearby merchants, no purchase event will occur, and further conversion will not be generated.

[0005] Therefore, the exposure of surrounding businesses in commercial operations is greatly restricted and difficult, and it is urgent to realize the active exposure of various surrounding businesses to outbound users. Summary of the Invention

[0006] One purpose of this application is to achieve active exposure of various businesses around travel stations to outbound users.

[0007] According to one aspect of an embodiment of the present application, a method for recommending surrounding businesses in a business operation is disclosed, the method comprising:

[0008] Determine the exit point where the outbound behavior is triggered based on the user's itinerary;

[0009] For the surrounding merchants radiating from the exit, estimate the conversion value of the products of the surrounding merchants to the user;

[0010] Based on the conversion value of the surrounding merchants to the user, the surrounding merchants of the outbound user are pushed.

[0011] According to one aspect of an embodiment of the present application, determining the exit to which the triggered travel behavior belongs based on the user itinerary generated by the user includes:

[0012] Acquire the user's travel data, wherein the travel data at least describes the user's current travel and the exit stations and exit gates involved in the user's historical travel;

[0013] For the user's travel data, the number of times the user has exited each gate in the gate group at the current station is obtained from the gate information according to the exit station and exit gate corresponding to the exit behavior triggered by the user;

[0014] The exit gate to which the triggered exit behavior belongs is determined according to the distribution of the number of times the user exits the station at each gate group at the current station.

[0015] According to one aspect of an embodiment of the present application, estimating the conversion value of the products of the surrounding merchants radiating from the exit to the user includes:

[0016] For each of the surrounding merchants radiating from the exit, the user's prediction of each product is performed to obtain the user's click rate for each product of the surrounding merchants;

[0017] According to the corresponding product value and the number of times the products of the surrounding merchants are exposed to the user, the conversion value of all the products of the surrounding merchants delivered to the user is calculated based on the click rate of each product of the surrounding merchants.

[0018] According to one aspect of an embodiment of the present application, estimating the click rate of each product of each surrounding merchant radiated by the exit to obtain the click rate of each product of the surrounding merchant by the user includes:

[0019] For each surrounding merchant radiating from the exit, a user-product pair is constructed based on the user profile of the user and the product features of each product of the surrounding merchant;

[0020] By predicting the product click-through rate of the user-product pair, the click-through rate of the user on each product of the surrounding merchants is obtained.

[0021] According to one aspect of an embodiment of the present application, the user portrait corresponds to the user's own travel data, and a combination of the user's own travel data and third-party supplementary data.

[0022] According to one aspect of an embodiment of the present application, the conversion value of all products of the surrounding merchants delivered to the user is calculated based on the click rate of each product of the surrounding merchants according to the corresponding product click value and the number of times the product of the surrounding merchants is exposed to the user, including:

[0023] The number of times the user's exit behavior is triggered and attributed to the exit is the number of times the user's products from the surrounding merchants are exposed to the user;

[0024] For the products of the surrounding merchants, the weighted sum of the click-through rates is calculated based on the corresponding product click value and the number of exposures of the user to obtain the conversion value of the product delivered to the user.

[0025] According to one aspect of an embodiment of the present application, the pushing of surrounding merchants to an outbound user based on the conversion value of the surrounding merchants to the user includes:

[0026] For the surrounding businesses radiating from the exit, determine the surrounding businesses with high conversion value based on the conversion value of each surrounding business to the user;

[0027] The products of the high-conversion-value surrounding merchants are pushed on the terminal page to which the outbound behavior triggered by the user jumps.

[0028] According to one aspect of an embodiment of the present application, a surrounding merchant recommendation system in commercial operations is disclosed, the system including a memory, a processor, and a computer program stored in the memory, the processor executing the computer program to implement the steps of the method as described above.

[0029] According to one aspect of an embodiment of the present application, a computer-readable storage medium is disclosed, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described above are implemented.

[0030] According to one aspect of an embodiment of the present application, a computer program product is disclosed, including a computer program, wherein the computer program is executed by a processor to implement the steps of the method described above.

[0031] In the embodiment of the present application, for each outbound user who triggers an outbound behavior, the exit port to which the triggered outbound behavior belongs is determined based on the user itinerary generated by the user, and the conversion value of the products of the surrounding merchants radiating from the exit port is estimated. The surrounding merchants of the outbound user are pushed based on the conversion value of the surrounding merchants to the user. In this way, the surrounding merchants are actively pushed to the outbound user, and the products of the pushed surrounding merchants are actively exposed. The exposure of the products of the surrounding merchants is no longer limited to the passive triggering of the user.

[0032] As surrounding merchants and their products are actively exposed to outbound users, and since the products actively exposed are based on the conversion value of the products to the outbound users, the exposed products are of interest to the outbound users. As the active exposure progresses, purchase events are very likely to occur, thereby generating value conversion of the products.

[0033] Other features and advantages of the present application will become apparent from the following detailed description, or may be learned in part by practice of the present application.

[0034] It should be understood that the foregoing general description and the following detailed description are merely illustrative and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The above and other objects, features and advantages of the present application will become more apparent by describing in detail example embodiments thereof with reference to the attached drawings.

[0036] Figure 1 A flowchart of a method for recommending surrounding businesses in commercial operations according to one embodiment of the present application is shown.

[0037] Figure 2 is based on Figure 1 The corresponding embodiment shows a method flow chart describing the step of determining the exit gate attribution of the triggered outbound behavior based on the user itinerary generated by the user.

[0038] Figure 3 It is a flow chart of a method for describing the steps of distributing the number of times a user exits each gate at the current station gate group according to an embodiment of the present application, and determining the exit in the exit direction as the exit to which the triggered exit behavior belongs.

[0039] Figure 4 It is a flow chart of a method for describing the steps of determining the exit to which the triggered exit behavior belongs based on the distribution of the number of times the user exits each gate at the current station gate group according to another embodiment of the present application.

[0040] Figure 5 is based on Figure 4 The corresponding embodiment shows a method flow chart describing the steps of predicting the exit gate to which the user belongs based on the distribution of the number of times the user exits the station at each gate group at the site and the user portrait.

[0041] Figure 6 is based on Figure 1 The corresponding embodiment shows a method flow chart describing the steps of estimating the conversion value of the products of surrounding merchants radiating from the exit to users.

[0042] Figure 7 is based on Figure 6 The corresponding embodiment shows a method flow chart describing the steps of performing user estimation on each product of each surrounding merchant radiating from the exit and obtaining the user's click rate on each product of the surrounding merchant.

[0043] Figure 8 is based on Figure 6The corresponding embodiment shows a method flow chart describing the steps of calculating the conversion value of all products delivered to users by surrounding merchants based on the click value of the corresponding product and the number of times the products of surrounding merchants are exposed to the user, and the click rate of each product of the surrounding merchants.

[0044] Figure 9 is based on Figure 1 The corresponding embodiment shows a method flow chart describing the steps of pushing surrounding merchants to outbound users based on the conversion value of the surrounding merchants to the user. DETAILED DESCRIPTION

[0045] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of this application will be more comprehensive and complete and will fully convey the concepts of the example embodiments to those skilled in the art. The accompanying drawings are merely schematic illustrations of the present application and are not necessarily drawn to scale. Identical reference numerals in the figures indicate identical or similar parts, and thus repeated descriptions thereof will be omitted.

[0046] In addition, the described features, structures or characteristics may be combined in one or more example embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the example embodiments of the present application. However, those skilled in the art will appreciate that the technical solutions of the present application may be practiced while omitting one or more of the specific details, or other methods, components, steps, etc. may be adopted. In other cases, known structures, methods, implementations or operations are not shown or described in detail to avoid obscuring the main content and making various aspects of the present application vague.

[0047] Some of the blocks shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0048] Travel stations are dotted with numerous businesses, such as those located near subways, buses, high-speed trains, airports, and ships. To this end, businesses operating in these stations act as surrounding merchants corresponding to a station exit, offering products to users exiting the station at that exit. If interested, these users may purchase the products, thereby achieving product value conversion.

[0049] If a user exiting a station, for example, arrives at the station and triggers exit at the exit gate, but does not visit nearby businesses or actively search for nearby businesses based on their location, the products of nearby businesses will not be exposed to the user. Furthermore, even if the user actively searches for nearby businesses, they are often unable to locate their location due to the lack of GPS (Global Positioning System) signals, making it impossible to search for nearby businesses based on their location.

[0050] It can be seen from this that in commercial operations around travel stations, there are many difficulties in exposing surrounding businesses and their products.

[0051] To this end, an embodiment of the present application provides a method for recommending surrounding businesses in commercial operations, so as to actively expose surrounding businesses and their products to outbound users, thereby enhancing the reliability and accuracy of geographic location-based applications in travel scenarios.

[0052] See Figure 1 , Figure 1 A flowchart of a method for recommending surrounding businesses in commercial operations according to one embodiment of the present application is shown.

[0053] An embodiment of the present application provides a method for recommending surrounding businesses in a commercial operation, the method comprising:

[0054] Step S110, determining the exit to which the exit behavior is triggered based on the user's itinerary;

[0055] Step S120: For the surrounding merchants radiating from the exit, estimate the conversion value of the products of the surrounding merchants to the user;

[0056] Step S130: Push surrounding merchants to the outbound user based on the conversion value of the surrounding merchants to the user.

[0057] These steps are described in detail below.

[0058] First of all, it should be noted that the surrounding merchant recommendations implemented by this application are made for each outbound user. In other words, through the embodiments of this application, the application based on geographic location will be actively implemented for each outbound user in a timely and accurate manner, that is, the surrounding merchants and their products will be actively exposed.

[0059] In step S110, as the user triggers the exit behavior, on the one hand, a user itinerary corresponding to the exit behavior will be generated; on the other hand, the triggering of the exit behavior will also trigger the active exposure of the merchants and their products around the corresponding exit to the user.

[0060] The triggering of exit behavior refers to the user passing through the exit gate, such as the triggering of exiting by scanning a code. The user who triggers the exit behavior will successfully pass through the exit gate and become an exit user.

[0061] Users who initiate an exit behavior and pass through the exit gate and successfully exit the station will generate corresponding travel data, namely the user itinerary. The user itinerary indicates the currently relevant information such as the exit station and exit gate, which can be used as a reference for the current geographical location to accurately determine the exit gate to which the user's exit behavior is triggered.

[0062] A user who triggers the exit behavior faces more than two passable exits. It is impossible to know which exit the user will pass through, and therefore it is certainly impossible to know which nearby merchant at the exit the user will arrive at will accept the delivery of his or her products.

[0063] During the execution of step S110 , as the outbound behavior is triggered, the exit to which the outbound behavior triggered by the user belongs will be determined based on the user itinerary generated.

[0064] As mentioned above, as the user's outbound behavior is triggered, step S110 will be executed, that is, the active exposure of surrounding merchants and their products implemented by this application will be initiated, so as to ensure the immediacy and time reliability of the active exposure implemented, and effectively avoid delayed recommendations.

[0065] For each user, the triggering of his / her exit behavior will also be used to trigger the active exposure of surrounding merchants and their products to the user as implemented in this application. In addition, the user's current exit will be accurately located based on the user's itinerary, and the user's exit passage will be located, so that the user's positioning at the site is no longer limited to the positioning of GPS signals.

[0066] Among them, the exit to which the outbound behavior triggered by the user belongs refers to the exit that the user currently passes through, as described in step S110, which is used as a basis to determine the user itinerary of the exit to which the currently triggered outbound behavior belongs, including the user itinerary currently generated by the user and the historical itinerary.

[0067] The user's trip is obtained from the user's trip data. In other words, the user's current trip and historical trips can be obtained from the user's trip data, and the exit station and exit gate involved can be obtained from it to determine the exit gate that triggered the current exit behavior.

[0068] It can be clearly seen that the exit positioning of the user achieved through step S110 does not require GPS positioning based on location permissions in the station hall and its surroundings. Under the action of step S110, the user's itinerary generated by the travel can be accurately mapped to the exit through which the user exits the station, solving the problem of positioning difficulty during travel, so that the user can actively initiate exposure of surrounding businesses and their products by adapting to the determined exit during travel.

[0069] It is further explained that in step S110, for the station where the user exits, the number of exits at each gate on the exit gate group is obtained from the travel data, that is, the number of exits at each gate of the exit gate group at the station by the user.

[0070] The number of times a user exits a station at each gate is calculated from their OriginDestination (OD) data. For example, the user's OD data indicates the user's trip generated when entering and exiting the gate, as well as the exit stations and exit gates involved in the generated user trip.

[0071] For each user trip generated by the user, it can be known that the user exits from a certain gate at a certain station. Therefore, by counting all the user trips of the user, the number of times the user exits from each gate at a station can be determined.

[0072] On this basis, combined with the existing gate information, the gate group of each gate to be counted, as well as the line and city where the gate group is located, are determined. Finally, the user obtains complete information on the number of exits from each gate in the station gate group.

[0073] In an exemplary embodiment, the travel data exists in the form of a user travel table. Each user travel record generated when the user enters or exits the station through the gate is stored in the constructed user travel table.

[0074] Among them, the field information in the user's itinerary includes: date, itinerary identification (trip id), user identification (user id), city identification (city id), line number, entry time, entry station code, entry gate code, exit time, exit station code and exit gate code, etc., which are not listed here one by one.

[0075] In one exemplary embodiment, gate information is collected based on site design and indicates the location of each gate. This information is stored in a gate information table, and the fields in the table include: gate code, city ID, route number, site code, gate group number within the station, gate station number within the station, and device type.

[0076] By counting the travel data and matching the travel data with the gate information, we can determine the number of times a user exits a station at a gate, as well as the location of the gate, and especially the gate group where the gate is located, so as to ultimately determine the number of times a user exits a station at each gate in the station gate group.

[0077] By executing step S110, the gates used to exit the station at each station through the gate group are recorded and counted, so as to match the user's travel trajectory at this station with the exit.

[0078] See also Figure 2 , Figure 2 is based on Figure 1 The corresponding embodiment shows a method flow chart describing the step of determining the exit gate attribution of the triggered outbound behavior based on the user itinerary generated by the user.

[0079] The step S110 of determining the exit to which the exit behavior is triggered based on the user's itinerary provided in the embodiment of the present application includes:

[0080] Step S111: Acquire the user's travel data, where the travel data at least describes the user's current travel and the exit stations and exit gates involved in the user's historical travel;

[0081] Step S112: Based on the user's trip data and the exit station and exit gate corresponding to the user's exit behavior, the user's exit times at each gate in the current gate group are obtained from the gate information;

[0082] Step S113, determining the exit gate to which the triggered exit behavior belongs based on the distribution of the number of times the user exits the station at each gate group at the current station.

[0083] These steps are described in detail below.

[0084] The travel data of the user who currently triggers the travel behavior is obtained. For example, the travel data is the user travel table referred to above. The user travel and the exit gate code in the user travel are obtained from the user travel table.

[0085] For the statistics of the number of times a user exits a station at a gate, the number of times the user exits the station at the corresponding gate is counted according to the exit gate code of each user trip, and the gate group where the gate is located is determined based on the gate information, so as to finally obtain the number of times the user exits the station at each gate in the station gate group.

[0086] For example, the gate information is the gate information table mentioned above. For each station, each gate group and each gate on the gate group are collected to build a relational database to obtain the gate information table.

[0087] To further explain, the gate information table represents the location of the gate, and the gate code it contains indicates the unique code of the gate in the city. For example, the gate code can be composed of the site code, equipment type and equipment number, and the equipment number is the gate station serial number in the gate information table.

[0088] For example, if the site code is 0253, the device type is 01, and the device number is 07, then the gate number is 02530107.

[0089] As mentioned above, in addition to the gate code, the fields in the gate information table also include city identification, line number, station code, gate group number within the station, gate station number, and device type.

[0090] The city ID is used to uniquely identify the city. In other words, each city has a unique city ID. The site code is used to uniquely identify the site in the city, so that each site has a unique code within the city. The gate group serial number within the station is used to identify each group of gates in the station. For example, if there are 4 rows of gates in the station, they correspond to serial numbers 1-4 respectively. The gate station serial number is the equipment number. For example, if there are 10 gates in the station, the equipment number can be serial numbers 1-10.

[0091] Based on the gate information table and user itinerary table, the user's exit count at each gate in the station gate group is obtained. This exit count describes the user's exit behavior. For users who generate user itineraries as they travel, the aforementioned steps are executed to obtain the exit count of the exit gates at the gate group corresponding to the destination station in their itinerary. This data is then used as the basis for accurately mapping the user's user itinerary to the corresponding station to the exit gate, enabling the user to locate the exit gate during the exit process.

[0092] Under the action of step S111 and step S112, the number of exits of the user at each gate group at a station is obtained, and then in the execution of step S113, the exit gate to which the user belongs is predicted based on the distribution of the number of exits, or the distribution of the number of exits and the corresponding user portrait, so as to determine the most likely exit gate for the user's user itinerary.

[0093] In one exemplary embodiment, the execution process of step S113 includes: analyzing the distribution of the number of times a user exits the station at each gate group at the current station, and determining the exit in the preferred exit direction as the exit to which the exit behavior is triggered. It should be noted that if a user frequently exits a gate group, it indicates that the gate corresponding to that gate group in that exit direction is the exit that the user prefers. For this user, the exit exit to which the user's current exit behavior is triggered can be determined by using the exit group's preferred exit statistics, thereby reducing computational complexity and improving execution efficiency.

[0094] See also Figure 3 , Figure 3 It is a flow chart of a method for describing the steps of distributing the number of times a user exits each gate at the current station gate group according to an embodiment of the present application, and determining the exit in the exit direction as the exit to which the triggered exit behavior belongs.

[0095] The embodiment of the present application provides a step of determining the exit in the exit direction as the exit to which the exit behavior is triggered based on the distribution of the number of exits of the user at each gate group at the station, including:

[0096] Step S301, determining the gate that the user prefers in the gate group based on the distribution statistics of the number of times the user exits the station at each gate group at the site;

[0097] Step S302: Determine the exit on the same side as the user's destination based on the location of the gate.

[0098] The following describes these two steps in detail.

[0099] For a gate group, statistics are performed on the distribution of the number of times a user exits the station at each gate in the gate group to determine whether the user tends to exit the station at the left half of the gate group or the right half of the gate group.

[0100] According to the statistical determination of the user's preference at the gate group, the exit direction and exit gate can be determined, and then the exit gate on the same side is the user's commonly used exit gate. This exit gate is used as the user's belonging exit gate, and the user's user itinerary and travel trajectory are matched with this exit gate.

[0101] Therefore, after determining the exit to which the user's current exit behavior is triggered, for the surrounding business recommendations implemented by this application, the user's travel trajectory can be mapped to the exit, thereby realizing trajectory tracking within the station hall, breaking away from the restricted GPS positioning and the required location permissions, and also locating the exit of the exiting user, greatly enhancing the tracking performance within the station.

[0102] Not limited to the statistics of the exit in the direction of exit, in another exemplary embodiment of the present application, the exit to which the user's current exit behavior is triggered can also be predicted in combination with the user's user portrait.

[0103] See also Figure 4 , Figure 4 It is a flow chart of a method for describing the steps of determining the exit to which the triggered exit behavior belongs based on the distribution of the number of times the user exits each gate at the current station gate group according to another embodiment of the present application.

[0104] The embodiment of the present application provides a step of determining the exit gate to which the triggered exit behavior belongs based on the distribution of the number of exits of the user at each gate group at the current station, including:

[0105] Step S401, obtaining a user profile of the user;

[0106] Step S402: predict the exit gate to which the user belongs based on the distribution of the number of times the user exits the station at each gate group at the site and the user portrait.

[0107] The following describes these two steps in detail.

[0108] The user profile is used to describe the user status of the user, which can be composed of constructed user tags. Exemplarily, the user profile exists in the form of a user tag table, which includes a user ID and several user IDs corresponding to the user ID.

[0109] The user ID will uniquely identify the user. User tags include but are not limited to gender, age, life stage, occupation, total number of rides in the past 7 / 14 / 30 / 90 days, total ride duration in the past 7 / 14 / 30 / 90 days, average arrival time of the first user trip on a weekday in the past 7 / 14 / 30 / 90 days, average departure time of the most recent trip on a weekday in the past 7 / 14 / 30 / 90 days, average commuting distance per trip in the past 7 / 14 / 30 / 90 days, total number of rides, high-frequency stops, total number of commercial benefits received, total number of commercial benefits write-offs, total amount of commercial benefits write-offs, average customer price of commercial benefits write-offs, total number of travel benefits received, number of each card voucher, and frequency of card use.

[0110] The distribution of the number of times a user exits each gate in a station gate group describes the number of times a single user exits gate 1 in the gate group, the number of times a single user exits gate 2 in the gate group, and so on, at gate n_zj in the gate group (n_zj is the number of gates in the entire gate group).

[0111] The distribution of the number of times a user exits the station at each gate group at the site and the user portrait are used as the characteristics of the user to predict the exit gate to which the user belongs.

[0112] It should be understood that for a user, the number of exits at a single group of gates within a certain time range, such as a single month, may be low, and affected by multiple factors such as actual conditions. Therefore, the distribution of the number of exits at each gate in the site gate group and the user portrait are used as the characteristics of the user, which provides a data basis for predicting the exit to which the user belongs, and also improves the richness and scalability of the features, which is conducive to achieving accurate predictions based on existing data.

[0113] The outbound gate to which the user belongs is predicted, and it should be noted that the outbound gate obtained by the prediction is an outbound gate corresponding to the gate group, and it is the outbound gate through which the user can pass after exiting from the gate group.

[0114] The gate group corresponds to at least one outbound gate, and the outbound gate through which the user is most likely to pass after exiting from a gate of the gate group is the outbound gate to which the user is predicted to belong.

[0115] The prediction of the outbound gate to which the user belongs can obtain the probability of the outbound gate to which the user belongs through a pre-trained classification model, and then determine the outbound gate to which the user belongs by the probability.

[0116] For example, the pre-trained can be a binary classification model, which takes the distribution of the number of times that the user exits from each gate of the gate group in the station and the user portrait as features, and labels the corresponding direction value of the feature according to the direction of the user after exiting under the feature. Each direction value uniquely corresponds to an outbound gate through which the user can pass after exiting from the gate group.

[0117] For example, if the user after exiting from the gate group passes to the left as mapped by the feature, the direction value is labeled as 0; and if the user after exiting from the gate group passes to the right as mapped by the feature, the direction value is labeled as 1.

[0118] Based on this, a training data set is constructed Each training data set D train in each training data set D is a feature of the record, which is used to describe the user state of the corresponding user.

[0119] Specifically, includes the distribution of the number of times that a user exits from each gate of the gate group in the station and the user portrait, and the user portrait can be represented by a plurality of user labels. is the feature of user i at gate group j, R m is an m-dimensional real number space of the m-dimensional feature mapping of user i.

[0120] For example, x ij ∈R m = [date (YYYYMM), user label 1 of user i, user label 2 of user i, …, user label n label of user i, number of times that user i exits from gate 1 of gate group j, number of times that user i exits from gate 2 of gate group j, …, number of times that user i exits from gate n zj of gate group j], date (YYYYMM) represents the month, such as 202312, which indicates December 2023, and the date feature indicates the time range corresponding to other features. And user label n labelRefers to the nth label User tags, that is, n r Represents the number of users, n i represents the number of gate groups that user i faces when exiting the station during travel, n label Represents the number of user tags, y ij ∈{0,1} is the direction value of the feature annotation in a single record of the corresponding training dataset. ij ∈{0,1} represents the final exit direction of user i under the corresponding characteristics (especially the actual number of exits).

[0121] That is, y in a single record of the training dataset ij is the marked direction value. For example, if user i turns left after exiting gate group j, then y ij =0, if going right, then y ij =1.

[0122] By training the model with the constructed training data set, a model is obtained that can accurately predict the user's exit, providing users with accurate judgment of their exit affiliation. There is no need to obtain location, that is, it is no longer restricted by GPS positioning and the acquisition of location permissions.

[0123] Furthermore, the training for exit prediction is performed on gate groups corresponding to a single exit. A training dataset is constructed for each gate group, and features are constructed based on all records related to that gate group. These features are annotated with direction values ​​based on the gate group's unique orientation.

[0124] Specifically, the number of times the user exits each gate of the gate group in the site is obtained through the travel data and gate information, that is, the user gate group exit data of user-site-gate group-gate-exit times is obtained, which can be in a tabular form, that is, a table maintained according to fields such as user, site, gate group, gate, and exit times, and the user gate group exit data corresponding only to the unique exit is obtained therefrom, and the number of times the user exits each gate of the gate group at the site is obtained from the user gate group exit data corresponding only to the unique exit, and the feature is constructed based on this.

[0125] In addition, the corresponding user portraits will be extracted for feature construction, and the constructed features will be labeled with direction values ​​to finally form a training data set.

[0126] The constructed training dataset shields the influence of other exits and only considers the prediction of the user's exit in the feature dimension, so that subsequent predictions will not be affected by other factors, which will greatly improve the accuracy and reliability of the prediction.

[0127] Exemplarily, a gate group entrance and exit distance relationship will also be constructed. The gate group entrance and exit distance relationship is used to indicate the distance of each gate of the gate group in a site relative to the exits in each direction. Therefore, after predicting the most likely direction for the user to exit the gate, it can be directly mapped to the closest exit based on the gate group entrance and exit distance relationship. This exit is the exit to which the predicted user belongs.

[0128] As previously mentioned, the model trained using the constructed training dataset can be a binary classification model. This binary classification model is then used to predict a user's assigned exit gate. Specifically, the binary classification model predicts the user's most likely exit direction from each gate group based on the user's exit count at each gate group and the user's profile. The exit gate mapped to this direction is then used as the user's assigned exit gate.

[0129] Further explanation: the number of times users exit each gate at the site gate group and the user profile are used to construct a prediction data set. Among them, unlike the training data set, the prediction data set is not limited to the gate group corresponding to the only exit; in other words, it will not distinguish whether the gate group corresponds to the only exit. All data of the user will be used to construct the prediction data set, and then the exit to which the user belongs will be predicted based on the features of the constructed prediction data set. predict In each record Represents the characteristics of user i in gate group j whose exit gate has not yet been determined. As mentioned above, this characteristic describes the number of exits of user i at each gate in gate group j and the user tag. The characteristic list included is the same as similar.

[0130] Specifically, to obtain the applicable two-class model model exit , first construct the problem loss function for the tth round of iteration:

[0131]

[0132] in, is the predicted value of the ijth record in the tth iteration, f t is the t-th tree, f t (x) = w q (x), q represents the tree f t The structure of w is the tree f t The weight of the leaf, and w i To represent the weight of the i-th leaf, Ω(f t ) is about tree f t The complexity of the regular term, then L (t) At point f t (xi ) Make a second-order approximation and get:

[0133]

[0134] in,

[0135]

[0136] So far, the optimal solution to the problem is:

[0137]

[0138] Where q is a tree with a fixed structure, γ and λ are the first-order and second-order regularization coefficients, respectively.

[0139] The optimal solution of each round based on the training data set can be iteratively converged to obtain a binary classification model exit .

[0140] Based on the obtained two-class model exit , the prediction data set D predict Each record of Use the two-classification model model one by one exit Implement the prediction and obtain the probability that the characteristics of user i in gate group j are biased in one direction, that is, the probability that the user i belongs to the exit corresponding to gate group j.

[0141] For example, if the gate group j corresponds to two exits. In other words, after exiting the station through the gate group j, user i has two directions, each direction corresponds to an exit. If the user i leans towards the other direction, the probability of him belonging to the other exit is

[0142] In addition, if there are more than two exits in one direction of gate group j, that is, the exit in this direction is not unique, then when the binary classification model predicts that user i tends to go in a direction with more than two exits after exiting gate group j, the exit that user i is most likely to pass through will be determined based on the traffic flow of each exit, and this exit is the exit to which user i belongs.

[0143] Specifically, each exit in one direction has its own flow rate, which can be detected by the passenger flow counter installed at the exit. The flow rate corresponding to each exit is used to determine the flow ratio of each exit, and then the exit to which user i belongs is determined based on the flow ratio.

[0144] To further clarify, determining the exit that user i is most likely to pass through based on the traffic volume at each exit is a process that randomly selects a number proportionally to determine the user i's affiliation. The proportion referred to here is the traffic volume proportion of each exit.

[0145] After the binary classification model predicts the direction that user i will take after exiting gate group j, if there are multiple exit gates in the desired exit direction, the exit gate to which user i belongs will be determined based on the traffic proportion of each exit gate and the random number taken for user i.

[0146] For multiple exits in the outbound direction, the random number ranges mapped to each are constructed based on the traffic proportion of each exit. For example, there are three exits in the same outbound direction, and the traffic proportions of each of these three exits are 0.3, 0.2, and 0.5 respectively. Thus, the random number range mapped to each exit is constructed, that is, the random number range constructed for the first exit with a traffic proportion of 0.3 is [0, 0.3), the random number range constructed for the second exit with a traffic proportion of 0.2 is [0.3, 0.3+0.2), and the random number range constructed for the third exit with a traffic proportion of 0.5 is [0.3+0.2, 0.3+0.2+0.5).

[0147] For user i, a random number between 0 and 1 is assigned to the exit corresponding to user i. If the random number falls within the range [0, 0.3), user i's journey corresponds to the first exit; if the random number falls within the range [0.3, 0.3+0.2), user i's journey corresponds to the second exit; if the random number falls within the range [0.3+0.2, 0.3+0.2+0.5), user i's journey corresponds to the third exit.

[0148] For further information, see Figure 5 , Figure 5 is based on Figure 4 The corresponding embodiment shows a method flow chart describing the steps of predicting the exit gate to which the user belongs based on the distribution of the number of times the user exits the station at each gate group at the site and the user portrait.

[0149] The embodiment of the present application provides step S402 of predicting the exit gate to which the user belongs based on the distribution of the number of times the user exits each gate at a subway station gate group and the user profile, including:

[0150] Step S4021: Predict the user's exit times at each gate at the station gate group and the user's profile to obtain the probability of the user's exit direction, where the exit direction corresponds to at least one exit gate;

[0151] Step S4022, determining the outbound port to which the user belongs according to the probability of the outbound direction of the user.

[0152] The two steps are described in detail below.

[0153] In the execution of step S4021, the features are constructed with the outbound times of the user at each gate of the gate group of the station and the user tags contained in the user portrait, and the prediction is performed through the pre-trained model, such as the binary classification model as indicated above, to obtain the probability of each outbound direction after the user exits the gate group of the designated station.

[0154] The directions through which the user can pass after exiting a gate group are more than one, for example, in many cases, the user will have one or two outbound directions after exiting a gate group, and there are passable outbound ports in each outbound direction.

[0155] After obtaining the probability of the outbound direction of the user, the direction through which the user passes after exiting the gate group can be determined according to the probability, that is, the outbound direction corresponding to the high probability.

[0156] The outbound port in the outbound direction corresponding to the high probability is determined. In an exemplary embodiment, in the case where there is only one outbound port in the outbound direction corresponding to the high probability, the outbound port closest to the obtained gate group exit-entry distance relationship is determined according to the outbound direction corresponding to the high probability, and this outbound port is taken as the outbound port to which the user belongs.

[0157] In another exemplary embodiment, there are more than two outbound ports in the outbound direction corresponding to the high probability, for example, the import and export channels in this outbound direction branch into more than two outbound ports, at this time, the outbound port to which the user belongs is determined from the more than two outbound ports according to the proportion of the flow rate of each outbound port and the random number obtained.

[0158] Therefore, it is no longer necessary to obtain the position of the user, that is, it is no longer necessary to obtain the position permission and then perform GPS positioning, and the user journey and user trajectory of the user can be accurately corresponded to the predicted outbound port, and accurate prediction of the user trajectory after the user exits is realized.

[0159] In another exemplary embodiment of the application, before step S301, the step S113 of determining the outbound port to which the triggered outbound behavior belongs according to the distribution of the outbound times of the user at each gate of the gate group of the current station further comprises:

[0160] determining whether the outbound times of the user at the gate group of the station are less than a set threshold, if the outbound times are less than the set threshold, performing steps S401 to S402, and if the outbound times are not less than the set threshold,

[0161] The step of distributing the number of times the user exits each gate at the current station gate group and determining the exit in the exit direction as the exit to which the triggered exit behavior belongs is executed.

[0162] The number of times a user exits a station gate group is used to determine whether the user frequently exits the station through the current station gate group. The number of times a user exits a station gate group can be obtained based on the user's travel data and gate information. Further, the number of times a user exits a station gate group can be obtained from the user's travel data and gate information. Of course, this is not limited to this. The number of times a user exits a station gate group can also be obtained by counting the user's exit data at each gate in the station gate group, which is not limited here.

[0163] For example, the set threshold value may be 5, that is, if the user exits a station through a gate group no less than 5 times within the set time range, it means that the exit direction corresponding to the gate group and the exit in the exit direction are consistent with the user's exit passage, thereby determining that the exit is the user's commonly used exit.

[0164] To further explain, for the number of exits that is not less than the set threshold, the gate machine that the user prefers will be respectively determined, that is, based on the distribution of the number of exits of the user in the gate machine group, the left half or the right half of the gate machine group that the user prefers will be counted. Regardless of the left half or the right half of the gate machine, it corresponds to an exit direction. The exit direction that the user will pass through after exiting the station can be determined by the gate machine to which the user prefers, and then the exit in the exit direction is the exit to which the user belongs.

[0165] At this point, it should be added that if the number of gates in a gate group is odd, the middle gate will belong to both the left and right half groups of gates. Therefore, for users exiting the station through the middle gate, it will be counted as the number of times the user exited the station through the left half group of gates and the number of times the user exited the station through the right half group of gates.

[0166] If the number of times a user exits a station gate group is less than the set threshold, it means that neither the station nor the gate group is frequently used by users. Therefore, it is necessary to construct features based on richer data, that is, the distribution of the number of times users exit each gate in the station gate group and the user portrait as mentioned above, and then implement predictions through a pre-built model.

[0167] In another exemplary embodiment, a sliding time window is set for exit gate determination based on user profiles, for example, a 30-day sliding window. This sliding time window is used to control the acquisition of the user's exit count at each gate in the station gate group, and even the acquisition of the user profile, thereby ensuring the validity of the obtained data and thereby enhancing the accuracy of the location corresponding to the user's itinerary.

[0168] Correspondingly, the user's travel data and gate information are also updated as the user's itinerary is added and the gate position changes. The travel data and gate information are controlled by a sliding time window to obtain the number of times the user exits the station at each gate group at the station.

[0169] In an exemplary embodiment, a distance factor is also set, which is used to measure the distance between the elevator facility from the platform to the station hall and the gate position.

[0170] That is, for the gate where the user has exited the station, determine whether the distance between the elevator facility and the gate position exceeds the distance factor. If it exceeds the distance factor, the exit to which the user belongs is determined by executing the above process.

[0171] If the distance between the elevator facility and the gate does not exceed the set distance factor, the user is determined to have chosen the gate closest to the elevator facility to exit, and the exit at the same location as the gate is the user's exit.

[0172] In this way, users' exit behavior can be screened, thereby shielding the impact of the distance between elevator facilities and gates.

[0173] After determining the exit point to which the user's outbound behavior belongs, we can use this as a basis to define the range of surrounding merchants recommended to the user, and then implement product recommendations within the defined range of surrounding merchants.

[0174] After the execution of step S110, the exit port to which the current exiting user belongs is determined, the execution of step S120 will determine the surrounding merchants radiating from the exit port, and then the corresponding conversion value of each surrounding merchant's product delivered to the current exiting user will be estimated, so as to obtain the conversion value of each surrounding merchant radiating from the exit port to the current exiting user.

[0175] During the execution of step S120, it should be understood that there are a number of businesses located near the exit, and these businesses are referred to as the businesses surrounding the exit. These businesses surrounding the exit can be located along the exit passage corresponding to the exit, outside the exit, or along the path from the track area within the station hall to the exit, without limitation.

[0176] Several surrounding merchants radiating towards the exit estimate the conversion value of their products to the current exit user one by one, obtain the conversion value of each product to the current exit user, and finally obtain the conversion value of all products of surrounding merchants to the current exit user.

[0177] For example, the conversion value that a product provides to a user indicates the user's interest in the product, and thus also indicates the possibility that the product can be purchased by the user, that is, quantifies the possibility of a purchase event occurring.

[0178] Correspondingly, the conversion value of all products from surrounding merchants to users refers to the degree of attention received by users after all products from surrounding merchants are released to users. It is the value of all products from surrounding merchants to a single user, which is used to measure the possibility of purchase events occurring for surrounding merchants and all products from surrounding merchants.

[0179] For each user leaving the station, after determining the exit and the surrounding merchants radiating from the exit, the delivery value directed to the user leaving the station can be estimated for all the products on sale and available at the surrounding merchants, that is, the conversion value shown in step S120.

[0180] In step S120 , the conversion value of the product delivered to the user is calculated using a pre-trained click model. For example, the conversion value of the product delivered to the user is estimated using a click-through rate prediction model to obtain the conversion value of each product for the user.

[0181] In other words, the CTR prediction model estimates the conversion value of multiple products for a single user, obtaining the conversion value of each product for that user. For a nearby merchant, the weighted sum of the conversion values ​​of each product for that user is calculated to obtain the conversion value of all products from that merchant for that user.

[0182] In an exemplary embodiment, a click-through rate prediction model for estimating the conversion value between users and products launched by merchants is trained through the click records of a large number of users on a travel application on a large number of products exposed.

[0183] Specifically, a training dataset is first constructed for the click-through rate prediction model. Each record in the training dataset includes a user profile of the user and product features constructed for the product. The product is exposed on the travel application and clicked by the user, that is, it corresponds to the user's click record on the exposed product on the travel application.

[0184] Thus, the training data set is constructed

[0185] Among them, the constructed training data set D train A single record in Right now A user portrait is composed of multiple data sources, including but not limited to travel data, application tracking data, etc.

[0186] Specifically,

[0187] corresponds to the user's own travel data, and This corresponds to third-party supplementary data. For example,

[0188] For product features, exemplary,

[0189] y ij ∈{0,1}, when the value is 0, it means that no click event occurs after product j is exposed to user i; when the value is 1, it means that a click event occurs after product j is exposed to user i.

[0190] As a result, travel applications have been widely used in the past d For each user i in the day, all n i Product generation records to form the training data set for the click-through rate prediction model.

[0191] Correspondingly, the conversion value estimate for users and products is obtained through a trained click-through rate prediction model. For this click-through rate prediction model, a corresponding prediction dataset must be constructed. Specifically, a prediction dataset is constructed for each user using user profiles and product features. Each record in the prediction dataset represents a constructed user-product pair.

[0192] The constructed prediction data set is Each record, that is, the user product pair The feature list of is the same as that of the training dataset.

[0193] is the number of optional products of the surrounding merchant s. Each user-product pair represents the recommendation of the surrounding merchant s to the user who passed by during period T. For example, if the user

[0194] Each record in the prediction dataset represents the recommendation of product j to a user, i.e., user 1, 2, ..., n user Each record will be used as the input of the click-through rate prediction model to obtain the user's click-through rate for the product. Similarly, the user's click-through rate for all products of surrounding merchants will be obtained, and finally the conversion value of surrounding merchants on the user will be obtained.

[0195] In another exemplary embodiment, for the click-through rate prediction model, the problem loss function constructed at the t-th iteration is:

[0196]

[0197] in, is the predicted value of the ijth record in round t, f t is the t-th tree, f t (x) = w q (x), q represents the tree f t The structure of w is the tree f t The weight of the leaf is w i To represent the weight of the i-th leaf, Ω(f t ) is about the tree f t The complexity of the regular term, then L (t) At the point Doing a second-order approximation, we get:

[0198]

[0199] in,

[0200]

[0201] The optimal solution is:

[0202]

[0203] In summary, the optimal solution of each round based on the training data set can be iteratively converged to obtain the click rate prediction model model nearby .

[0204] The prediction data set D predict Each record of Use click-through rate estimation model one by one nearby Implementation estimates.

[0205] Finally, the click rate, click value, and user exposure times are calculated. Perform weighted summation to obtain the conversion value of the nearby merchant s to user i through the nearby merchant function of the travel application. is the number of products available for selection by surrounding merchants, ctv j is the conversion value of the corresponding product, is the number of exposures of user i in period T, and is also the number of times user i passes by the surrounding merchant s in period T.

[0206] See also Figure 6 , Figure 6 is based on Figure 1The corresponding embodiment shows a method flow chart describing the steps of estimating the conversion value of the products of surrounding merchants radiating from the exit to users.

[0207] The embodiment of the present application provides a step S120 of estimating the conversion value of the products of surrounding merchants radiating from the exit to the user, including:

[0208] Step S121: For each of the surrounding businesses radiating from the exit, the user's prediction of each product is performed to obtain the user's click rate for each product of the surrounding businesses;

[0209] Step S122 , based on the corresponding product value and the number of times the products of the surrounding merchants are exposed to the user, the conversion value of all the products of the surrounding merchants delivered to the user is calculated based on the click rate of each product of the surrounding merchants.

[0210] Among them, it should be explained first that the commodity value referred to is the click value of the corresponding commodity. For example, it can be the contribution rate of the purchase event caused by the commodity being put into the user, or it can be the valuation of the commodity, which is not limited here.

[0211] The number of times a user is exposed to products from neighboring merchants is the number of times the user passes by this neighboring merchant within the period (cycle) T. The number of times a user is exposed to products from neighboring merchants can affect whether the user purchases the products from this neighboring merchant.

[0212] In the CTR estimation, the user and each product from the surrounding merchants form a user-product pair, which is used to estimate the user's click-through rate for each product from the surrounding merchants. The CTR is used to indicate the user's appeal to the product; the higher the CTR, the more attractive the product is to the user.

[0213] For the user-product pair consisting of a user and a product, the click-through rate of the user on the product is estimated using the click-through rate prediction model as described above, thereby representing the conversion value of the product to the user. Through such recommendations, the click-through rate of all products in the surrounding stores delivered to the user is obtained.

[0214] Users have corresponding click-through rates for all products, which can numerically describe the possibility of users purchasing products.

[0215] After obtaining the user's click rate for each product, that is, the conversion value of each product delivered to the user, it is accumulated through weighted summation to ultimately obtain the conversion value of all products delivered to the user by surrounding merchants.

[0216] See also Figure 7 , Figure 7 is based on Figure 6 The corresponding embodiment shows a method flow chart describing the steps of performing user estimation on each product of each surrounding merchant radiating from the exit and obtaining the user's click rate on each product of the surrounding merchant.

[0217] The embodiment of the present application provides step S121 of performing user predictions on each product of each surrounding merchant radiating from the exit, and obtaining the user's click-through rate of each product of the surrounding merchant, including:

[0218] Step S1211: For each surrounding merchant radiating from the exit, a user-product pair is constructed based on the user's user profile and the product features of each product of the surrounding merchant;

[0219] Step S1212: Obtain the click rate of the user on each product of the surrounding merchants by predicting the click rate of the product of the user-product pair.

[0220] The following describes these two steps in detail.

[0221] First of all, it should be explained that users generate corresponding user itineraries as they initiate entry and exit behaviors. The user itineraries of numerous users and each user will constitute itinerary data. That is to say, users will exist in the form of user itineraries in the data. Therefore, it is possible to construct a user profile describing the user's travel based on this, and thus characterize the user with the user profile for use in building user-product pairs with the products to be launched.

[0222] The surrounding businesses recommended to outbound users belong to the outbound user's current exit. Therefore, the user's itinerary used to construct the word portrait is related to this exit and the station where this exit is located, thereby ensuring the accuracy of the features of the constructed user-product pairs and eliminating useless or even interfering data.

[0223] For merchants around the site, outbound applications that can respond to users' outbound behavior are deployed to each user to actively implement recommendations for surrounding merchants and products on sale in response to outbound behavior, without the need for users to passively trigger it.

[0224] Therefore, each user only needs to interact with the gate through the travel application running on their user terminal to initiate their own entry or exit behavior, such as scanning the QR code provided by the travel application on the gate to enter or exit the station. The travel application running on the user terminal and the system corresponding to the travel application will obtain the user's own travel data, such as the user's itinerary.

[0225] In other words, the user's own travel data will be used to represent the user in the data to build the user's user profile. The user's own travel data is the multi-part data owned by the deployed travel application, such as the user's itinerary data mentioned above.

[0226] In addition, users' own travel data also includes user information on travel applications, embedded data on travel applications, statistical data based on travel data, users' commercial operation data on travel applications, user-related product data, etc., which are not listed here one by one.

[0227] For example, user tracking data on travel apps is used to describe the user's clicks on exposed products; user information on travel apps includes gender, age, region, and occupation; statistical data based on travel data may specifically include, but is not limited to, the total number of rides in the past 77 / 14 / 30 / 90 days, the total ride duration in the past 7 / 14 / 30 / 90 days, the average arrival time of the first trip on weekdays in the past 7 / 14 / 30 / 90 days, the average departure time of the last trip on weekdays in the past 7 / 14 / 30 / 90 days, the average commuting distance in the past 7 / 14 / 30 / 90 days, the total number of rides, and the top 10 most frequent stops;

[0228] User business operation data on travel apps, including but not limited to: total number of business benefits redeemed, total number of business benefits redeemed, total amount of business benefits redeemed, average customer order value of business benefits redeemed, total number of travel benefits redeemed, number of card vouchers redeemed, and card usage frequency;

[0229] User-related product data may include but is not limited to: the top 10 product categories viewed in the past 1 / 3 / 7 / 14 / 30 / 90 days, the corresponding number of times the top 10 product categories were viewed in the past 1 / 3 / 7 / 14 / 30 / 90 days, and the average price of the top 10 product categories viewed in the past 1 / 3 / 7 / 14 / 30 / 90 days).

[0230] It should be understood that the user's own travel data will be adapted to the travel application and provide data for the recommendation of surrounding businesses, so that the surrounding business recommendations can obtain the corresponding user portrait.

[0231] The source of user profiles is not limited to the user's own travel data, but can also come from third-party supplementary data. In short, user profiles correspond to the user's own travel data, as well as a combination of own travel data and third-party supplementary data.

[0232] Third-party supplementary data includes, but is not limited to, whether the individual is pregnant or having children, the degree of gaming interest, whether the individual is budget-conscious, whether the individual is a high-net-worth individual, whether the individual frequently flies, whether the individual is interested in financial management, whether the individual is interested in the anime, whether the individual is budget-conscious, whether the individual is a "phone addict", whether the individual likes to go shopping in supermarkets, whether the individual likes to take the children out for a walk, whether the individual is willing to buy a car, whether the individual has a hospital to buy a house, whether the individual likes to visit museums, whether the individual likes to visit historical and cultural attractions, whether the individual likes to visit natural scenic areas, and other data.

[0233] For the user whose exit is determined by the current triggered exit behavior, the user's user profile is obtained through the user's own travel data, or a combination of the user's own travel data and third-party supplementary data, and then the user profile is used as the user feature in the constructed user-product pair.

[0234] In other words, the user portrait digitized by the user's own travel data, or the combination of the user's own travel data and third-party supplementary data, will serve as the user feature and constitute a user-product pair together with the product features of the product.

[0235] The product features of a product are used to describe the product. For example, the product features of a product include descriptive features, statistical features, and statistical generalization features of the product.

[0236] Specifically, the product characteristics include the first / second / third-level category to which the product belongs, the price of the product, the number of exposures of the product in the past 1 / 3 / 7 / 14 / 30, the number of clicks on the product in the past 1 / 3 / 7 / 14 / 30, the number of purchase conversions of the product in the past 1 / 3 / 7 / 14 / 30, the number of days the product has been on the shelf, the number of exposures of the third-level category to which the product belongs in the past 1 / 3 / 7 / 14 / 30, the number of clicks on the third-level category to which the product belongs in the past 1 / 3 / 7 / 14 / 30, and the number of purchase conversions of the third-level category to which the product belongs in the past 1 / 3 / 7 / 14 / 30.

[0237] For each product of a nearby merchant, a user-product pair is constructed using the user characteristics and product characteristics of the current outbound user, thereby predicting the user's click-through rate on the product based on the user-product pair.

[0238] It should be understood that when surrounding merchants place goods on display to users, users are attracted by the exposed goods, which can be compared to clicks on goods exposed on the Internet. Therefore, the degree of attraction of the goods exposed by surrounding merchants to users can be accurately measured by users' clicks on the goods.

[0239] The pre-built click prediction model is used to predict the click-through rate (CTR) of user-product pairs, thereby obtaining the user's CTR. It should be understood that the CTR of a user's product is the probability that the user will be attracted to the product, or the likelihood that the user will be attracted to the product after it has been exposed by surrounding merchants.

[0240] See also Figure 8 , Figure 8 is based on Figure 6 The corresponding embodiment shows a method flow chart describing the steps of calculating the conversion value of all products delivered to users by surrounding merchants based on the click value of the corresponding product and the number of times the products of surrounding merchants are exposed to the user, and the click rate of each product of the surrounding merchants.

[0241] The embodiment of the present application provides a step S122 of calculating the conversion value of all products of surrounding merchants delivered to the user based on the click rate of each product of the surrounding merchants according to the corresponding product click value and the number of times the products of the surrounding merchants are exposed to the user, including:

[0242] Step S1221: The number of times the user's exit behavior is triggered and attributed to the exit is the number of times the user's products from surrounding merchants are exposed to the user;

[0243] In step S1222, for the products of the surrounding merchants, the weighted sum of the click-through rate is calculated based on the corresponding product click value and the number of user exposures to obtain the conversion value of the product delivered to the user.

[0244] By counting the user's outbound behaviors attributed to the current exit during the user's itinerary, the number of outbound behaviors triggered by the user attributed to the current exit is obtained, which is used as the number of times the user's products from surrounding merchants are exposed to the user.

[0245] Therefore, the conversion value of the surrounding merchant s relative to the user i can be estimated, that is:

[0246]

[0247] in, The number of products available for nearby merchants, ctv j is the click value of the products in the surrounding merchants s, is the number of times user i is exposed to surrounding merchants s during period T.

[0248] At this point, the execution of step S120 is completed, and then the push of the recommended surrounding merchants and their products on the user's terminal page can be achieved under the action of step S130.

[0249] See also Figure 9 , Figure 9 is based on Figure 1 The corresponding embodiment shows a method flow chart describing the steps of pushing surrounding merchants to outbound users based on the conversion value of the surrounding merchants to the user.

[0250] The step S130 of recommending surrounding merchants to the outbound user based on the conversion value of the surrounding merchants to the user provided in the embodiment of the present application includes:

[0251] Step S131: For the surrounding businesses radiating from the exit, determine the surrounding businesses with high conversion value based on the conversion value of each surrounding business to the user;

[0252] Step S132: Push products from nearby merchants with high conversion value to the terminal page to which the user triggers the outbound behavior.

[0253] For several surrounding merchants radiating from the exit, based on the conversion value of each surrounding merchant to the current exit user, the surrounding merchants with high conversion value are directly selected for the current exit user, and then the surrounding merchants with high conversion value are pushed to the current exit user.

[0254] As the terminal page of the travel application run by the current outbound user responds to the triggered outbound behavior, it will receive push notifications from surrounding businesses with high conversion value, allowing the recommendations of surrounding businesses to be proactively made under the triggering of the outbound behavior, thereby effectively improving the activity of the ecology around the site.

[0255] For businesses around the site, with the active recommendation of surrounding businesses implemented by this application, the traffic generated by outbound users is transmitted to surrounding businesses, greatly enhancing the traffic monetization ability of travel applications.

[0256] The following is a specific example to illustrate the surrounding merchant recommendations implemented by this application.

[0257] For each nearby merchant, a user-product pair is constructed based on the product and the current outbound user, and then the click rate of the current outbound user on the product is obtained.

[0258] Specifically, the user's itinerary table and gate information table are used to obtain the distribution of the number of times a user exits a station at a gate group. If a user exits a station by scanning a QR code using a travel app, the distribution of the number of times the user scans the code at each gate group describes the number of times the user scans the code at each gate, thus obtaining the user's gate group scan table.

[0259] That is to say, the user's travel table is associated with the gate information table through the gate code to obtain the user's gate group code table.

[0260] Based on the user gate group code swiping table, the distribution of the user's code swiping at this group of gates can be known, and the distribution of the user's code swiping at this group of gates will indicate the user's exit direction.

[0261] Combined with the gate group entrance and exit distance relationship described in the gate group entrance and exit distance relationship table, the exits to which the gate group points in different directions can be known, and the user's frequently used exits can be determined, and the user's recently frequently used exit table can be obtained.

[0262] For each exit, the surrounding businesses mapped to the exit are recorded to build and maintain the surrounding business exit relationship table.

[0263] By mapping the exits to the surrounding businesses, we can know the surrounding businesses that the user will pass by when leaving the station from an exit.

[0264] For example, the frequently used exit gates can be determined by counting the user's preferred exit directions in the user gate group swiping code table. In addition, if the user has fewer user trips in the user trip table, for example, the number of exits of the user at the station gate group is less than the set threshold, it will be necessary to determine the user's preferred exit direction through the distribution of the number of exits of the user at each gate of the station gate group and the user portrait through a binary classification model, and then determine the exit gate to which the user belongs in this exit direction.

[0265] A pre-trained binary classification model is used to determine the probability of the user belonging to each exit, and the probability is then used to determine the user's exit. For example, the pre-trained binary classification model can be characterized by the distribution of the number of times a user exits each gate at the station and the user's profile. This feature is annotated with a corresponding direction value based on the direction the user travels after exiting the station. Each direction value uniquely corresponds to an exit accessible from the gate.

[0266] For example, if the user in the feature mapping passes to the left after exiting the gate group, the marked direction value is 0; correspondingly, if the user in the feature mapping passes to the right after exiting the gate group, the marked direction value is 1.

[0267] Based on this, construct the training data set Among them, each training data set D train In a single record It is the feature of this record, which is used to describe the user status of the corresponding user.

[0268] Specifically, It includes the distribution of the number of times a user exits each gate in a site gate group at a certain time and the user portrait, and the user portrait can be represented by multiple user tags. is the characteristic of user i in gate group j, R m is the m-dimensional feature of user i.

[0269] For example, x ij ∈R m= [date (YYYYMM), user label 1 of user i, user label 2 of user i, …, user label n of user i label , number of times of user i exiting from gate 1 of gate group j, number of times of user i exiting from gate 2 of gate group j, …, number of times of user i exiting from gate n of gate group j zj ], date (YYYYMM) represents a month, for example, 202312, which indicates December 2023, and the date feature indicates the time range corresponding to other features. User label n label refers to the n label th user label of user i, n r represents the number of users, n i represents the number of gate groups facing the user i in the exit, n label represents the number of user labels, y ij ∈{0,1} is the direction value labeled in the single record of the corresponding training data set. y ij ∈{0,1} represents the direction of the final exit passage of user i corresponding to the feature (especially in the actual number of exits).

[0270] That is, y ij in the single record of the training data set is the labeled direction value. For example, if user i exits gate group j to the left, then y ij = 0, and if to the right, then y ij = 1.

[0271] Through model training based on the constructed training data set, a model capable of accurately predicting the exit port to which the user belongs is obtained, thereby providing accurate judgment of the exit port belonging to the user, and no longer needing to obtain the location, i.e., no longer being limited by GPS positioning and location permission acquisition.

[0272] Further, the training for exit port prediction will be performed on the gate group corresponding to only the unique exit port. The construction of the training data set is performed on the gate group corresponding to the unique exit port, and the features are constructed according to all records related to the gate group, and the direction values of the constructed features are labeled according to the unique direction of the gate group.

[0273] Specifically, the number of times of user i exiting from each gate of the gate group in the station is obtained through the travel data and gate information, i.e., user gate group exit data of user-station-gate group-gate-exit times is obtained, which can be in the form of a table, i.e., a table maintained according to the fields of user, station, gate group, gate, and exit times, and the user gate group exit data corresponding to only the unique exit port is obtained therefrom, the number of times of user i exiting from each gate of the gate group in the station is obtained from the user gate group exit data corresponding to only the unique exit port, and the features are constructed therefrom.

[0274] In addition, the corresponding user portraits will be extracted for feature construction, and the constructed features will be labeled with direction values ​​to finally form a training data set.

[0275] The constructed training dataset shields the influence of other exits and only considers the prediction of the user's exit in the feature dimension, so that subsequent predictions will not be affected by other factors, which will greatly improve the accuracy and reliability of the prediction.

[0276] Exemplarily, a gate group entrance and exit distance relationship will also be constructed. The gate group entrance and exit distance relationship is used to indicate the distance of each gate of the gate group in a site relative to the exits in each direction. Therefore, after predicting the most likely direction for the user to exit the gate, it can be directly mapped to the closest exit based on the gate group entrance and exit distance relationship. This exit is the exit to which the predicted user belongs.

[0277] As previously mentioned, the model trained using the constructed training dataset can be a binary classification model. This binary classification model is then used to predict a user's assigned exit gate. Specifically, the binary classification model predicts the user's most likely exit direction from each gate group based on the user's exit count at each gate group and the user's profile. The exit gate mapped to this direction is then used as the user's assigned exit gate.

[0278] Further explanation: the number of times users exit each gate at the site gate group and the user profile are used to construct a prediction data set. Among them, unlike the training dataset, the prediction dataset is not limited to the gate group corresponding to a unique exit; in other words, it will not distinguish whether the gate group corresponds to a unique exit. All user data will be used to construct the prediction dataset, and then the features of the constructed prediction dataset will be used to predict the exit to which the user belongs.

[0279] Prediction dataset D predict In each record Represents the characteristics of user i in gate group j whose exit gate has not yet been determined. As mentioned above, this characteristic describes the number of exits of user i at each gate in gate group j and the user tag. The characteristic list included is the same as similar.

[0280] Based on the obtained two-class model, the prediction dataset D predict Each record of Use the binary classification model to implement predictions one by one, and obtain the probability that the characteristics of user i in gate group j are biased in one direction, that is, the probability that the user i belongs to the exit corresponding to gate group j.

[0281] For example, if the gate group j corresponds to two outbound ports. In other words, the user i has two directions after the gate group j, and each direction corresponds to an outbound port, then the probability of the user i being inclined to another direction, i.e. belonging to another outbound port, is

[0282] In addition, if there are more than two outbound ports in a direction of the gate group j, i.e. the outbound ports in the direction are not unique, when the user i is predicted to be inclined to the direction with more than two outbound ports by the binary classification model, the outbound port through which the user i is most likely to pass is determined according to the flow of each outbound port, and the outbound port is the outbound port to which the user i belongs.

[0283] Specifically, each outbound port in a direction has its flow, which can be perceived by a passenger flow counter arranged at the outbound port. The flow proportion of each outbound port is determined according to the flow corresponding to each outbound port, and the outbound port to which the user i belongs is determined according to the flow proportion, and finally the recommendation for the surrounding businesses of the outbound port is realized.

[0284] Through the above description of the embodiments, those skilled in the art can easily understand that the example embodiments described herein can be implemented by software or by software combined with necessary hardware. Therefore, the technical solutions according to the embodiments of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a U disk, a mobile hard disk, etc.) or a network, and includes a plurality of instructions to make a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) execute the method according to the embodiments of the present application.

[0285] In the example embodiments of the present application, a computer program medium is also provided, which stores computer readable instructions, and when the computer readable instructions are executed by a processor of a computer, the computer executes the method described in the method embodiment part.

[0286] According to one embodiment of the present application, a program product for implementing the method in the above method embodiment is also provided, which can be in the form of a portable compact disc read-only memory (CD-ROM) and includes program code, and can be run on a terminal device, such as a personal computer. However, the program product of the present application is not limited to this, and in this document, the readable storage medium can be any tangible medium containing or storing a program, which can be used or combined with an instruction execution system, device or apparatus.

[0287] The program product may be implemented in any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or component, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0288] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium that can transmit, propagate, or transfer a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0289] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0290] The program code for performing the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, and the like, as well as conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a stand-alone software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device may be connected to the user computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0291] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiment of the application, the features and functions of two or more modules or units described above can be concretized in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be concretized.

[0292] Furthermore, although the steps of the method of the present application are described in a particular order in the accompanying drawings, this does not require or imply that the steps must be performed in this particular order, or that all steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps.

[0293] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solution according to the embodiments of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the embodiments of the present application.

[0294] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered merely as exemplary, and the true scope and spirit of the present application are indicated by the appended claims.

Claims

1. A method for recommending surrounding businesses in commercial operations, characterized in that: The method comprises: Acquire the user's travel data, wherein the travel data at least describes the user's current travel and the exit stations and exit gates involved in the user's historical travel; For the user's travel data, the number of times the user has exited each gate in the gate group at the current station is obtained from the gate information according to the exit station and exit gate corresponding to the exit behavior triggered by the user; Determine the exit gate to which the triggered exit behavior belongs based on the distribution of the number of exits of the user at each gate group at the current station; For each of the surrounding merchants radiating from the exit, the user's prediction of each product is performed to obtain the user's click rate for each product of the surrounding merchants; Calculate the conversion value of all products of the surrounding merchants delivered to the user based on the click-through rate of each product of the surrounding merchants according to the corresponding product value and the number of times the user is exposed to the products of the surrounding merchants; Based on the conversion value of the surrounding merchants to the user, the surrounding merchants of the outbound user are pushed.

2. The method according to claim 1, characterized in that The step of estimating the click rate of each product of each surrounding merchant radiated by the exit to obtain the click rate of each product of the surrounding merchant by the user includes: For each surrounding merchant radiating from the exit, a user-product pair is constructed based on the user profile of the user and the product features of each product of the surrounding merchant; The click rate of each product of the surrounding merchants by the user is obtained by predicting the product click rate of the user-product pair.

3. The method according to claim 2, characterized in that The user profile corresponds to the user's own travel data, and a combination of the user's own travel data and third-party supplementary data.

4. The method according to claim 1, wherein The conversion value of all products of the surrounding merchants delivered to the user is calculated based on the click rate of each product of the surrounding merchants according to the corresponding product click value and the number of times the user is exposed to the products of the surrounding merchants, including: The number of times the user's exit behavior is attributed to the exit is the number of times the user's products from the surrounding merchants are exposed to the user; For the products of the surrounding merchants, the weighted sum of the click-through rates is calculated based on the corresponding product click value and the number of exposures of the user to obtain the conversion value of the product delivered to the user.

5. The method according to claim 1, wherein The pushing of surrounding merchants to outbound users based on the conversion value of the surrounding merchants to the user includes: For the surrounding merchants radiating from the exit, determine the surrounding merchants with high conversion value based on the conversion value of each surrounding merchant to the user; The products of the high-conversion-value surrounding merchants are pushed on the terminal page to which the outbound behavior triggered by the user jumps.

6. A system for recommending nearby businesses in commercial operations, the system comprising a memory, a processor, and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 5.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

8. A computer program product comprising a computer program, characterized in that The computer program is executed by a processor to implement the steps of the method according to any one of claims 1 to 5.

Images