Store arrival prediction method, device, readable storage medium and electronic device

By extracting user and store features and entering into the store probability prediction model, the problem of inaccurate user positioning in searches near O2O is solved, and the accuracy and user experience of merchant screening results are improved.

CN110472995BActive Publication Date: 2024-11-01HANHAI INFORMATION TECH SHANGHAI
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN201910610494.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2019-07-08
Publication Date
2024-11-01
Estimated Expiration
2039-07-08

AI Technical Summary

Technical Problem

When the existing O2O near-point search function is inaccurate, it will lead to deviations in merchant screening results, affecting user decision-making and experience, especially in large buildings, and it is difficult to ensure the positioning accuracy.

Method used

By obtaining the search request from the user terminal, extracting the characteristics of the user and the store, crossing the characteristics, inputting them into the store probability prediction model, predicting the probability of the user reaching the store, and determining the precise positioning of the user's current location.

Benefits of technology

It improves the accuracy of the store probability prediction model and the accuracy of the target store, enhances the reliability of the user terminal recommendations to users, and improves the user experience.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN110472995B_ABST
    Figure CN110472995B_ABST
Patent Text Reader

Abstract

This application provides a store arrival prediction method, apparatus, readable storage medium, and electronic device. The method includes: obtaining a search request initiated by a user terminal; extracting features from the search request to determine the user's characteristics, the characteristics of each store in the store list corresponding to the search request, and user-store cross-features; inputting the user's characteristics, the characteristics of each store, and the user-store cross-features into a pre-trained store arrival probability prediction model to determine the probability that the user will arrive at each store; and determining whether the user has arrived at a target store based on the probability that the user has arrived at each store. This application determines the target store that the user will arrive at by using the probability of each store output by the store arrival probability prediction model, achieving accurate positioning of the user's current location and improving the reliability of stores near the user's current location recommended by the user terminal.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of information processing technology, and in particular to a store arrival prediction method, device, readable storage medium and electronic device. Background Art

[0002] With the development of mobile Internet, people can easily access the Internet through mobile devices to obtain services, thus a number of O2O (Online-to-Offline) local life services (such as O2O nearby search) have emerged, which greatly facilitates people's lives. Taking O2O nearby search as an example, users can use this search function to view food, entertainment and other life information near their current location. The specific implementation process of this search function is: pre-collect the longitude and latitude annotations of each merchant, obtain the user's location in real time, calculate the distance between the user's location and each merchant with annotated longitude and latitude, sort the merchants whose distance meets the user's screening range according to a certain algorithm to obtain the screening results, and return the screening results to the mobile device used by the user for display.

[0003] However, there is a problem in the specific implementation process of the above search function: the screening results are overly dependent on the user's location. When the user's location does not reach a certain accuracy, there is a deviation between the calculated user's location and the distance between each merchant, resulting in some merchants being unable to be used as screening results and returned to the user, seriously affecting the user's decision-making and usage experience.

[0004] Therefore, in order to ensure that the calculated distance between the user's location and the distance between the merchants with marked longitude and latitude meet a certain accuracy, it is necessary to ensure that the user's location meets a high precision requirement. In actual situations, the user's real-time location is greatly affected by the network conditions. When the user initiates an O2O nearby search in some large buildings (for example, in a shopping mall), the ordinary positioning technology cannot locate the floor information of large buildings, and the mobile device cannot receive GPS signals indoors. The user's real-time location can only be achieved by relying on the base station of the network operator. The real-time location of the user obtained by relying on the base station of the network operator has drift phenomenon, and it cannot be guaranteed that the user's real-time location indoors meets the high precision requirements.

[0005] Therefore, how to locate the current position of the user more accurately is a problem that needs to be solved urgently in this field. Summary of the invention

[0006] The embodiments of the present application provide a store arrival prediction method, device, readable storage medium and electronic device, which can predict in real time whether a user has arrived at a store, thereby achieving accurate positioning of the user's current location.

[0007] A first aspect of an embodiment of the present application provides a method for predicting store arrivals, the method comprising:

[0008] Obtaining a search request initiated by a user terminal;

[0009] Extracting features from the search request to determine features of a user of the user terminal, features of each store in the store list corresponding to the search request, and user-store cross features, where the user-store cross features are obtained by cross-feature analysis of the features of the user and the features of each store;

[0010] Inputting the characteristics of the user, the characteristics of each store, and the user-store cross-features into a pre-trained store arrival probability prediction model to determine the probability of the user arriving at each store;

[0011] According to the probability of the user arriving at each of the stores, it is determined whether the user arrives at a target store, where the target store is one of the stores.

[0012] Optionally, the step of determining whether the user has arrived at a target store according to the probability of the user arriving at each store includes:

[0013] When the probability that the user arrives at the target store is greater than a preset probability threshold, determining that the user arrives at the target store; or

[0014] When the probability that the user arrives at the target store is greater than a preset probability threshold and the parameter value corresponding to the search request is within a preset effective parameter value range, it is determined that the user arrives at the target store.

[0015] Optionally, the method further comprises:

[0016] Determine the types of user behaviors associated with user visits to your store;

[0017] Extracting a first category of search records and a second category of search records from the search log of the user terminal, wherein the first category of search records are search records that conform to the user behavior type, and the second category of search records are search records whose corresponding search moments have a time difference with the search moments of the first category of search records within a preset time length;

[0018] Marking the search records that match the user behavior type in the first category of search records and the second category of search records as positive samples, and marking the search records that do not match the user behavior type in the second category of search records as negative samples;

[0019] The preset model is trained according to the positive samples and the negative samples to obtain the store visit probability prediction model.

[0020] Optionally, the step of training a preset model according to the positive samples and the negative samples to obtain the store arrival probability prediction model includes:

[0021] Perform feature extraction on the positive samples and the negative samples respectively, determine the features of the sample users corresponding to the positive samples and the negative samples respectively, the features of each sample store in the store list corresponding to the positive samples and the negative samples respectively, and the sample user-sample store cross-features, where the sample user-sample store cross-features are obtained by cross-feature analysis of the features of the sample users and the features of each sample store;

[0022] The characteristics of the sample users, the characteristics of the sample stores, and the cross-characteristics of the sample users and sample stores are used as training samples to train the preset model and obtain the store visit probability prediction model.

[0023] Optionally, after the step of determining that the user has arrived at the target store, the method further includes:

[0024] Extracting, from the search log of the user terminal, a search record corresponding to the search time after it is determined that the user has arrived at the target store;

[0025] In the case where the extracted search record is a search record for the target store, marking the extracted search record as a positive sample and increasing the weight of the positive sample;

[0026] If the extracted search record is not a search request for the target store, marking the extracted search record as a negative sample and reducing the weight of the negative sample; and

[0027] The store arrival probability prediction model is updated according to the positive samples after the weight is increased and the negative samples after the weight is reduced.

[0028] Optionally, the user's characteristics include at least one of the following: the name of the WIFI scanned or connected to by the user terminal and the corresponding signal strength, the device type of the user terminal, the longitude and latitude of the user terminal, the IP address of the user terminal, the user's user portrait, and the user's consumption preferences.

[0029] Optionally, the characteristics of each store include at least one of the following: the logo of each store, the WIFI name of each store, the average WIFI connection or scanning strength of each store, the longitude and latitude of each store, the category to which each store belongs, the price range of the goods sold in each store, the click-through rate of each store, and the visit rate of each store.

[0030] Optionally, the user-store cross feature is obtained by at least one of the following methods:

[0031] Determining the straight-line distance between the user terminal and each store according to the longitude and latitude of the user terminal and the longitude and latitude of each store;

[0032] Performing feature cross-talk on the signal strength of the WIFI of the store scanned or connected by the user terminal and the average connection or scanning strength of the WIFI of the store scanned or connected by the user terminal;

[0033] Performing feature intersection on the signal strength of the WIFI of the store scanned or connected by the user terminal and the straight-line distance between the user terminal and the store scanned or connected by the user terminal; and / or

[0034] A feature cross-talk is performed between the user click or consumption price of the user terminal and the per capita price of each store.

[0035] A second aspect of an embodiment of the present application provides a store arrival prediction device, the device comprising:

[0036] An acquisition module, used to obtain a search request initiated by a user terminal;

[0037] A feature extraction module, configured to extract features from the search request, determine features of the user of the user terminal, features of each store in the store list corresponding to the search request, and user-store cross features, wherein the user-store cross features are obtained by cross-feature analysis of the features of the user and the features of each store;

[0038] A probability prediction module, used to input the characteristics of the user, the characteristics of each store, and the user-store cross-features into a pre-trained store arrival probability prediction model to determine the probability of the user arriving at each store; and

[0039] A determination module is used to determine whether the user has arrived at a target store according to the probability of the user arriving at each store, and the target store is one of the stores.

[0040] Optionally, the determining module includes:

[0041] A first determining module is used to determine that the user has arrived at the target store when the probability that the user has arrived at the target store is greater than a preset probability threshold; or

[0042] The second determination module is used to determine that the user has arrived at the target store when the probability that the user has arrived at the target store is greater than a preset probability threshold and the parameter value corresponding to the search request is within a preset effective parameter value range.

[0043] Optionally, the device further comprises:

[0044] A third determination module is used to determine the user behavior type associated with the user visiting the store;

[0045] A first extraction module is used to extract a first type of search record and a second type of search record from the search log of the user terminal, wherein the first type of search record is a search record that conforms to the user behavior type, and the second type of search record is a search record whose corresponding search time has a time difference with the search time of the first type of search record within a preset time length;

[0046] a marking module, configured to mark the search records in the first category of search records and the second category of search records that match the user behavior type as positive samples, and mark the search records in the second category of search records that do not match the user behavior type as negative samples;

[0047] The training module is used to train the preset model according to the positive samples and the negative samples to obtain the store arrival probability prediction model.

[0048] Optionally, the training module includes:

[0049] A feature extraction submodule is used to extract features from the positive samples and the negative samples respectively, and determine features of sample users corresponding to the positive samples and the negative samples, features of each sample store in the store list corresponding to the positive samples and the negative samples, and sample user-sample store cross features, where the sample user-sample store cross features are obtained by cross-feature analysis of the features of the sample users and the features of each sample store;

[0050] The training submodule is used to train the preset model using the characteristics of the sample users, the characteristics of each sample store, and the cross-features of the sample users and sample stores as training samples to obtain the store arrival probability prediction model.

[0051] Optionally, the device further comprises:

[0052] A second extraction module is used to extract, from the search log of the user terminal, the search record at the corresponding search time after it is determined that the user has arrived at the target store;

[0053] a first weight adjustment module, configured to mark the extracted search record as a positive sample and increase the weight of the positive sample when the extracted search record is a search record for the target store;

[0054] a second weight adjustment module, configured to mark the extracted search record as a negative sample and reduce the weight of the negative sample when the extracted search record is not a search record for the target store; and

[0055] The updating module is used to update the store arrival probability prediction model according to the positive samples with increased weights and the negative samples with reduced weights.

[0056] Optionally, the user's characteristics include at least one of the following: the name of the WIFI scanned or connected to by the user terminal and the corresponding signal strength, the device type of the user terminal, the longitude and latitude of the user terminal, the IP address of the user terminal, the user's user portrait, and the user's consumption preferences.

[0057] Optionally, the characteristics of each store include at least one of the following: the logo of each store, the WIFI name of each store, the average WIFI connection or scanning strength of each store, the longitude and latitude of each store, the category to which each store belongs, the price range of the goods sold in each store, the click-through rate of each store, and the visit rate of each store.

[0058] Optionally, the user-store cross feature is obtained by at least one of the following methods:

[0059] Determining the straight-line distance between the user terminal and each store according to the longitude and latitude of the user terminal and the longitude and latitude of each store;

[0060] Performing feature cross-talk on the signal strength of the WIFI of the store scanned or connected by the user terminal and the average connection or scanning strength of the WIFI of the store scanned or connected by the user terminal;

[0061] Performing feature intersection on the signal strength of the WIFI of the store scanned or connected by the user terminal and the straight-line distance between the user terminal and the store scanned or connected by the user terminal; and / or

[0062] A feature cross-talk is performed between the user click or consumption price of the user terminal and the per capita price of each store.

[0063] A third aspect of an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the method described in the first aspect of the present application.

[0064] A fourth aspect of an embodiment of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executed, implements the steps of the method described in the first aspect of the present application.

[0065] A store arrival prediction method provided by an embodiment of the present application is adopted. First, a search request initiated by a user is obtained, and then feature extraction is performed on the search request. The extracted features (including: features of the user of the user terminal, features of each store in the store list corresponding to the search request, and user-store cross features) are input into the store probability prediction model to obtain the probability of the user arriving at each store in the store list, and finally the target store that the user arrives at is predicted based on these probabilities. The present application performs feature extraction on the search request initiated by the user terminal from three dimensions: user dimension, store dimension, and user-store cross dimension, which improves the accuracy of the results output by the store arrival probability prediction model and the accuracy of the target store finally predicted. In addition, the target store that the user arrives at is determined by the probability of each store output by the store arrival probability prediction model, which realizes the accurate positioning of the user's current location, improves the reliability of the stores near the current location recommended by the user terminal to the user, and enhances the user's experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments of the present application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0067] Figure 1 is a schematic diagram of the interaction between the user terminal and the backend server provided in various embodiments of the present application;

[0068] Figure 2 is a flow chart of a store arrival prediction method shown in an embodiment of the present application;

[0069] Figure 3 is a flow chart of obtaining user-store cross features shown in an embodiment of the present application;

[0070] Figure 4 is a flowchart of a method for training a store arrival probability prediction model shown in an embodiment of the present application;

[0071] Figure 5 is a flowchart of another method for training a store arrival probability prediction model according to an embodiment of the present application;

[0072] Figure 6 is a flowchart of a method for updating a store arrival probability prediction model shown in an embodiment of the present application;

[0073] Figure 7 It is a schematic diagram of a store arrival prediction device shown in one embodiment of the present application. DETAILED DESCRIPTION

[0074] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0075] Before describing each embodiment of the present application, the related technology is described first. In order to obtain a more accurate positioning of the user, in the related technology, a method of correcting the obtained user positioning is adopted, for example, one of the correction methods is: looking for a reference source within a certain range of the user's current position, for example: if the user's current network status is connected to WIFI, the coordinates of the WIFI can be used as a reference source, and then the user's positioning is corrected; for example: when the user is inside a large building (for example: inside a large shopping mall), because the GPS signal is blocked, the user's positioning can be corrected according to the surrounding geomagnetic information of the user's historical positioning.

[0076] However, the above two methods of correcting the obtained user positioning require obtaining corresponding data information, which is costly. For example, when the coordinates of WIFI are used as a reference source, it is necessary to additionally obtain the coordinate information of each WIFI in the area where the user is located. When the user's positioning is corrected using geomagnetic information, it is necessary to additionally obtain the geomagnetic information in the area where the user is located.

[0077] In the related art, another way to obtain the user's location is to collect the WIFI of the store to which the user is connected, the time the user stays in the store, and the user's purchase information in the store, and then formulate corresponding rules in combination with the business, and then determine whether the user is in the store based on the rules. However, this method only considers the WIFI of the store to which the user is connected, the time the user stays in the store, and the user's purchase information in the store, that is, it only determines whether the user is in the store from the user's side. In addition, this solution is overly dependent on whether the store has WIFI and whether the user has purchase behavior, and it does not have strong universality. Therefore, the accuracy of the positioning results obtained by this positioning method for user positioning is still not high.

[0078] In order to improve the accuracy of user positioning, the embodiment of the present application utilizes the fact that many businesses in O2O (Online-to-Offline) local life services can only be completed in merchants' stores, obtains users' search requests in real time and marks the consumption data of users' typical behaviors towards stores (for example, participating in group purchase and coupon verification activities in stores, purchasing flash discounts, signing in, uploading UGC, etc.), and then extracts features from multiple dimensions such as users, stores, and intersections between users and stores for these consumption data. The extracted features are then input into a pre-trained store arrival probability prediction model to obtain the probability of users arriving at each store, and finally, based on these probabilities, the target store that the user actually arrives at is obtained.

[0079] A store arrival prediction method provided in an embodiment of the present application is described in detail below.

[0080] Figure 1 Schematic diagram of the interaction between the user terminal and the backend server provided in each embodiment of the present application. Figure 1 , the backend server communicates with one or more user terminals (for example: Figure 1 The user terminal 1 to the user terminal n) in the communication connection are communicated to realize communication interaction. The background server can be a network server, a database server, etc. The user terminal can be a personal computer (PC), a tablet computer, a smart phone, etc.

[0081] Figure 2 is a flow chart of a method for predicting store arrivals shown in an embodiment of the present application, and the method is applied to Figure 1 The backend server in . Figure 2 The store arrival prediction method provided in an embodiment of the present application includes the following steps:

[0082] Step S11: Obtain a search request initiated by a user terminal.

[0083] In this embodiment, the user can input a search request in the user terminal, and the backend server is used to receive the search request sent by the user terminal. The user terminal is installed with terminal application software that can support the nearby search function, such as search application software, shopping application software, or application software that can provide other services to the user. Through such application software, the user can search for any target within the surrounding area of ​​the current location. After receiving the search request input by the user, the user terminal sends the search request to the backend server, and then receives the search results returned by the backend server for the search request and displays the search results to the user in the form of a store list on the page.

[0084] After receiving the user's search request, the user terminal actively sends the user data for this search request to the background server. Alternatively, after receiving the user's search request, the user terminal sends the user data for this search request to the background server the next time the background server requests the user data from the user terminal. Of course, other methods can also be used to send the user data for the search request to the background server, and the various embodiments of the present application do not impose specific restrictions on this.

[0085] Among them, user data includes user behavior data, user information and device status information of the user terminal. User behavior data refers to: data dynamically generated when a user performs user behavior on any store in the store list. User behavior can be general user behavior, such as click, favorite, share, etc.; it can also be behavior related to the store (that is, behavior that can only be completed after arriving at the store), such as: flash payment, group purchase coupon verification, self-service ordering, queuing, uploading UGC, user sign-in, etc., of course, it can also be other types of user behavior. This application includes but is not limited to the various user behaviors listed above. User information refers to: user portrait (for example: user's age, occupation, gender, etc.), user's consumption preferences (for example: categories of frequently purchased goods, price consumption range) and other information that characterizes the user's personal characteristics. The device status information of the user terminal refers to: the name of the WIFI scanned or connected to by the user terminal and the corresponding signal strength, the device type of the user terminal, the latitude and longitude of the user terminal, the IP address of the user terminal, etc.

[0086] For example, the backend server is the backend server of a certain consumer APP. Accordingly, the terminal application software installed on the user terminal that supports the nearby search function is a consumer APP that communicates and interacts with the backend server. When the user enters "hot pot" in the search bar of the consumer APP, a page showing multiple stores related to "hot pot" will pop up on the consumer APP. The user can click on any store to query the relevant information of the store, participate in flash purchases, group purchase coupons, user sign-in and other activities in the store. The consumer APP records the user behavior data generated by the user for this search request in real time, obtains user information through the user's personal account, and obtains the device status information of the user terminal through the application in the user terminal. The user behavior data, user information and device status information of the user terminal are used to generate user data for this search request and send it to the backend server.

[0087] Step S12: Perform feature extraction on the search request to determine the features of the user of the user terminal, the features of each store in the store list corresponding to the search request, and the user-store cross-features. The user-store cross-features are obtained by performing feature cross-feature extraction on the features of the user and the features of each store.

[0088] In various embodiments of the present application, feature extraction is performed on the search request from three dimensions: user dimension, store dimension, and cross dimension of user and store. Feature extraction is performed on the search request, specifically: feature extraction is performed on user data corresponding to the search request.

[0089] Specifically, determining the characteristics of the user of the user terminal means: extracting data that conforms to the user dimension from the user data as the characteristics of the user of the user terminal. For example, the characteristics of the user of the user terminal can be: the name of the WIFI scanned or connected by the user terminal and the corresponding signal strength, the device type of the user terminal, the latitude and longitude of the user terminal, the IP address of the user terminal, the user portrait of the user, and the consumption preference of the user. Of course, the characteristics of the user can also include other characteristics that can characterize the personality of the user. This application includes but is not limited to the characteristics of the user listed above.

[0090] The backend server pre-stores the information of all stores registered on the terminal application software. For example, when the terminal application software is a consumer APP, the backend server pre-stores the information of all stores registered on the consumer APP. After the user terminal initiates a search request, the backend server returns a list of stores that meet the search conditions to the user terminal. Therefore, determining the characteristics of each store in the store list corresponding to the search request means extracting the characteristics of each store in the store list from the pre-stored information of all registered stores. The characteristics of each store in the store list can be: the logo of each store, the WIFI name of each store, the average WIFI connection or scanning strength of each store, the longitude and latitude of each store, the category to which each store belongs, the price range of the goods sold in each store, the click rate of each store, and the visit rate of each store. Specifically, the category to which each store belongs represents the type of goods sold by each store. For example, if a store sells clothing, then the category to which the store belongs is clothing. For another example, if a store sells snacks or beverages, then the category to which the store belongs is catering. The visit-to-purchase rate of each store refers to the percentage of customers who make purchases among all customers who visit the store within a certain period of time.

[0091] User-store cross features are the association features between users and each store. Figure 3 FIG. 1 is a flowchart of obtaining user-store cross-features according to an embodiment of the present application. Figure 3 , the user-store cross-features can be obtained through the following steps:

[0092] Step S121: determining the straight-line distance between the user terminal and each store according to the longitude and latitude of the user terminal and the longitude and latitude of each store.

[0093] Step S122: performing feature cross-talk on the signal strength of the WIFI of the store scanned or connected by the user terminal and the average connection or scanning strength of the WIFI of the store scanned or connected by the user terminal.

[0094] Performing feature cross-feature on feature A and feature B means: using a preset calculation method to calculate feature A and feature B to obtain an associated feature C of feature A and feature B. For example, feature A is the signal strength of the WIFI of the store scanned or connected to by the user terminal, and feature B is the average connection or scanning strength of the WIFI of the store scanned or connected to by the user terminal. The preset calculation method is the ratio of feature A to feature B, and the associated feature C is the value obtained by dividing feature A by feature B.

[0095] Step S123: Perform feature cross-talk on the signal strength of the WIFI of the store scanned or connected by the user terminal and the straight-line distance between the user terminal and the store scanned or connected by the user terminal; and / or perform feature cross-talk on the user click or consumption price of the user terminal and the per capita price of each store.

[0096] The per capita price of each store is the per capita consumption price of each store, the user click price is the price of the product clicked by the user, and the user consumption price is the price of the product purchased by the user. In order to ensure the accuracy of the target store finally predicted, this embodiment can perform feature crossover by any of the following three methods after performing feature crossover on the signal strength of the WIFI of the store scanned or connected by the user terminal and the average connection or scanning strength of the WIFI of the store scanned or connected by the user terminal:

[0097] 1) The signal strength of the WIFI of the store scanned or connected by the user terminal is feature-crossed with the straight-line distance between the user terminal and the store scanned or connected by the user terminal.

[0098] 2) Perform feature cross-talk between the user click price (or user consumption price) of the user terminal and the per capita price of each store.

[0099] 3) The signal strength of the WIFI of the store scanned or connected by the user terminal is cross-featured with the straight-line distance between the user terminal and the store scanned or connected by the user terminal, and the user click price (or user consumption price) of the user terminal is cross-featured with the per capita price of each store.

[0100] Step S13: Input the characteristics of the user, the characteristics of each store, and the user-store cross-features into a pre-trained store arrival probability prediction model to determine the probability of the user arriving at each store.

[0101] The store arrival probability prediction model is obtained by training a preset model using the search logs of the user terminal. The specific training process will be described in detail below.

[0102] After the user's features, the features of each store, and the user-store cross features are input into the store arrival probability prediction model, the store arrival probability prediction model will output the probability value of the user arriving at each store in the store list. The probability value of the user arriving at each store is proportional to the possibility of the user arriving at the store. The higher the probability value, the greater the possibility of the user arriving at the store, and the lower the probability value, the less likely the user is to arrive at the store.

[0103] Step S14: determining whether the user has arrived at a target store based on the probability of the user arriving at each store, where the target store is one of the stores.

[0104] The target store is the store where the user is currently located. Determining the target store means determining the user's current precise location. The specific steps for determining whether the user has reached the target store will be described below.

[0105] In one embodiment, the backend server is a backend server of a certain consumer APP, and the software installed on the user terminal that supports the nearby search function is a consumer APP that communicates and interacts with the backend server. When the user is inside a large shopping mall and searches for nearby stores through the consumer APP, the backend server returns multiple stores that meet the search conditions (the search conditions can be set from multiple aspects, such as: the distance from the user's current location, the price range of the store, and the user's favorable rating ranking) according to the user's search request for the user to view. For example, the user enters "hot pot" in the search bar and sets the search condition to be within a range of one kilometer from the current location. The consumer APP displays all stores related to "hot pot" that meet the search conditions on the page, and the user can view the store of interest or perform user behaviors related to a certain store (as described above, user behaviors can include: general user behaviors and behaviors related to visiting the store). The consumer APP records the user behavior data generated by the user in real time, and sends the user behavior data, user information, and device status information of the user terminal as user data for the user's current search request (search request with the search name "hot pot") to the backend server. The backend server extracts features from the user data and inputs the extracted features into the store arrival probability prediction model to obtain the probability of the user arriving at each store related to "hot pot" in the search results, and then predicts the store where the user is currently located based on each probability. In various embodiments of the present application, the current time is a short period of time including the time when the user initiates the search request.

[0106] In an embodiment of the present application, the search request initiated by the user is first obtained, and then the search request is feature extracted, and the extracted features (including: the features of the user of the user terminal, the features of each store in the store list corresponding to the search request, and the user-store cross-features) are input into the store probability prediction model to obtain the probability of the user arriving at each store in the store list, and finally the target store to be reached by the user is predicted based on these probabilities. The present application performs feature extraction on the search request initiated by the user terminal from three dimensions: user dimension, store dimension, and user-store cross-dimension, which improves the accuracy of the results output by the store probability prediction model and the accuracy of the target store finally predicted. In addition, the target store to be reached by the user is determined by the probability of each store output by the store probability prediction model, which realizes the positioning of the user's current location, improves the reliability of the stores near the current location recommended by the user terminal to the user, and enhances the user's experience.

[0107] Specifically, step S14 may include:

[0108] Step S141: when the probability that the user arrives at the target store is greater than a preset probability threshold, determining that the user arrives at the target store; or

[0109] Step S142: When the probability that the user arrives at the target store is greater than a preset probability threshold, and the parameter value corresponding to the search request is within a preset effective parameter value range, it is determined that the user arrives at the target store.

[0110] In this embodiment, there are two ways to determine whether the user has arrived at the target store. The first determination method is: compare the probability output by the store arrival probability prediction model with the preset probability threshold, and take the store with a probability greater than the preset probability threshold as the target store arrived by the user; the second determination method is: among the stores with a probability greater than the preset probability threshold, the store whose parameter value corresponding to the search request is within the preset effective parameter value range is taken as the target store arrived by the user, and the relationship between the parameter value corresponding to the search request and the preset effective parameter value range is used to assist in determining whether the user has arrived at the target store.

[0111] For example, if the preset probability threshold is 0.8, and among all the probabilities output by the store arrival probability prediction model, only one store M has a probability of 0.9, then according to the first determination method, store M is the target store that the user arrives at; according to the second determination method, if store M is a catering store, the parameter value corresponding to the search request can be set to the current time of the user terminal, and the preset effective parameter value range can be set to lunch time (for example: 11:00-13:00). If the current time of the user terminal happens to fall within the preset effective parameter value range, it can be determined that the user has arrived at store M. Of course, the parameter value corresponding to the search request and the preset effective parameter value range can be determined according to the needs in the actual application process of this application.

[0112] In actual situations, if the backend server obtains the search request initiated by the user terminal for a long period, there may be a phenomenon that the probability values ​​of multiple stores output by the store arrival probability prediction model are greater than the preset threshold. If the backend server obtains the search request initiated by the user terminal for a short period, the number of stores that meet the probability value greater than the preset threshold is usually small, or even zero. In this case, if there is only one store with a probability value greater than the preset threshold, the store is used as the target store where the user is currently located. If there are multiple stores with probability values ​​greater than the preset threshold, the store with the highest probability value among the multiple stores is used as the store where the user is currently located. The preset threshold is an empirical threshold for whether the user's store arrival judgment is effective, obtained by calculating the accuracy and recall rate of the store arrival probability prediction model, that is, the preset threshold is the target store predicted by the store arrival probability prediction model, and the user's actual store is compared and analyzed, and the optimal probability value is obtained in combination with the specific business. Only under this optimal probability value, the accuracy of the predicted target store to be reached by the user is the highest.

[0113] In an embodiment of the present application, two methods are set up to determine whether the user has arrived at the target store, which enhances the flexibility of the store arrival prediction method of the present application in actual application. In addition, the second determination method adds an auxiliary determination condition of whether the parameter value corresponding to the search request is within the preset effective parameter value range, thereby improving the accuracy of the determination result.

[0114] The following is an explanation of the training process of the store visit probability prediction model.

[0115] Figure 4 1 is a flowchart of a method for training a store arrival probability prediction model according to an embodiment of the present application. Figure 4 , the training method includes:

[0116] Step S21: Determine the user behavior type associated with the user's visit to the store.

[0117] User behaviors associated with user visits to a store refer to user behaviors that can only be completed after the user arrives at the store, such as flash discount purchases, group purchase coupon verification, self-service ordering, number queuing, uploading UGC, user sign-in, connecting to WIFI, etc. For a store, if a user generates user behaviors related to visiting the store, then it is more likely that the user actually arrives at the store.

[0118] Step S22: extracting first-category search records and second-category search records from the search log of the user terminal, wherein the first-category search records are search records that conform to the user behavior type, and the second-category search records are search records whose corresponding search time and the time difference between the search time of the first-category search record are within a preset time length.

[0119] The search log of the user terminal contains all search records. A search request initiated by a user can correspond to multiple search records. The multiple search records generated each time a search request is initiated constitute the user behavior data of the search request. For example, if the user enters "clothing" in the search bar, the search request with the search name "clothing" is search request X. After clicking the search, the user terminal will display multiple stores that match the search name "Chinese food". If user 1 visits store A and completes the check-in in store A, then user 1-search request X-store A-check-in is a search record. If user 1 visits store A and participates in the group purchase coupon verification activity in store A, then user 1-search request X-store A-group purchase coupon verification is another search record. Similarly, user 1 can also generate multiple search records based on other stores.

[0120] Among them, the first type of search record is a search record that meets the user behavior type. As long as the user behavior carried in the search record is a user behavior that can only be generated after the user arrives at the corresponding store, the search record can be regarded as a first type of search record. For example: a search record is user 1-search request X-store A-group purchase coupon verification. Since group purchase coupon verification requires the user to arrive at the corresponding store before it can be completed, this search record can be regarded as a first type of search record. For another example: a search record is user 1-search request X-store A-sharing. Since sharing does not require the user to arrive at the corresponding store, this search record cannot be regarded as a first type of search record.

[0121] Optionally, after selecting multiple first-category search records, further filtering conditions may be set. For example, the filtering condition may be a time range condition, and first-category search records whose corresponding search times in all search records are within a certain period of time are filtered out as new first-category search records.

[0122] In this embodiment, the first type of search records are only part of the search records in the search log that meet the user behavior type. In order to make the distribution of the collected search records reasonable, the second type of search records need to be collected. The second type of search records are search records whose time difference between the corresponding search time and the search time of the first type of search records is within the preset time length. For example, the search time of a first type of search record is 10:00, if the preset time length is 1 minute, then the second type of search record obtained from the first type of search record can be the search record generated by the user between 9:59-10:01.

[0123] The method of extracting the first category search records and the second category search records adopted in the present embodiment ensures that the extracted search records have a good distribution (for example: pre-extracting the first category search records with a more reasonable time distribution, and then extracting the second category search records based on these first category search records), avoiding the phenomenon that a certain number of search records that meet the user behavior type cannot be extracted, or the phenomenon that the time distribution of the extracted search records is unreasonable (for example: the search records are overly concentrated in a certain time period).

[0124] Step S23: Mark the search records in the first category of search records and the second category of search records that match the user behavior type as positive samples, and mark the search records in the second category of search records that do not match the user behavior type as negative samples.

[0125] A positive sample indicates that the user theoretically arrived at the store, and a negative sample indicates that the user theoretically did not arrive at the store. Since all user behaviors in the first category of search records are user behaviors associated with the user's visit to the store, all first category search records are positive samples. Since the first category of search records is only a part of the search records that meet the user behavior type in the search log, there may be multiple search records that meet the user behavior type in the second category of search records. Therefore, when dividing positive and negative samples, the search records that meet the user behavior type in the second category of search records need to be marked as positive samples, and the search records that do not meet the user behavior type need to be marked as negative samples.

[0126] Step S24: training a preset model based on the positive samples and the negative samples to obtain the store arrival probability prediction model.

[0127] In this embodiment, the positive samples and negative samples obtained by the division are used as input values, and a machine learning algorithm (for example, a logistic regression algorithm) is used to train the preset model to obtain a binary classification prediction model (i.e., a store arrival probability prediction model). When the features extracted from the search request are input, the probability of the user arriving at each store in the store list can be output. The store list is the search result corresponding to the search request.

[0128] Figure 5 FIG. 1 is a flowchart of another method for training a store arrival probability prediction model according to an embodiment of the present application. Figure 5 , step S24 comprises:

[0129] Step S241: Perform feature extraction on the positive samples and the negative samples respectively, determine the features of the sample users corresponding to the positive samples and the negative samples respectively, the features of each sample store in the store list corresponding to the positive samples and the negative samples respectively, and the sample user-sample store cross-features, wherein the sample user-sample store cross-features are obtained by performing feature cross-feature extraction on the features of the sample users and the features of each sample store.

[0130] Step S242: The preset model is trained using the characteristics of the sample users, the characteristics of the sample stores, and the cross-features of the sample users and sample stores as training samples to obtain the store visit probability prediction model.

[0131] In this embodiment, each search record also carries user information and device status information of the user terminal. After obtaining positive samples and negative samples, the characteristics of the corresponding sample user, the characteristics of the corresponding sample store, and the cross-features of the sample user and the sample store can be extracted from each sample. These characteristics are then input into the preset model and the preset model is trained to obtain a store visit probability prediction model.

[0132] Figure 6 FIG6 is a flowchart of a method for updating a store arrival probability prediction model according to an embodiment of the present application. Referring to FIG6 , the method includes:

[0133] Step S31: extracting the search record corresponding to the search time after it is determined that the user has arrived at the target store from the search log of the user terminal.

[0134] After the target store that the user will arrive at is predicted according to the store arrival probability prediction model, the search records of the corresponding search time after the user arrives at the target store are extracted from the search log of the user terminal, and these search records are used as feedback records. The feedback records can be used to adjust the sample weights, which is convenient for updating the store arrival probability prediction model. For example, if the user actually only arrives at store A, but may have made user behaviors related to the user's store visit to other stores, the samples related to store A can be marked as positive samples and the weight values ​​can be adjusted, and the samples of other stores can be marked as positive and negative and the weight values ​​can be adjusted to ensure the reliability of each sample when used.

[0135] Step S32: when the extracted search record is a search record for the target store, mark the extracted search record as a positive sample and increase the weight of the positive sample.

[0136] For example, if a prediction shows that the target store that the user will arrive at is store A, then in the feedback record, the search record related to store A can be used as a positive sample with a larger weight value. For example, when the user behavior in the search record related to store A is a group purchase coupon verification behavior (it can also be other user behaviors, such as uploading UGC, connecting to WIFI), the search record can be marked as a positive sample with a larger weight value.

[0137] Optionally, according to different user behaviors in the feedback records, this embodiment sets different weight values ​​for different positive samples (search records related to the target store in the feedback records). For example, among the positive samples, the weight value corresponding to the sample in which the user behavior is group purchase coupon verification behavior is greater than the weight value corresponding to the sample in which the user behavior is click behavior.

[0138] Step S33: When the extracted search record is not a search record for the target store, the extracted search record is marked as a negative sample, and the weight of the negative sample is reduced.

[0139] In this embodiment, the weight of search records that are not related to store A can also be adjusted. Specifically, the search records that are not related to store A are marked as negative samples, and the weight of the negative samples is reduced. For example, when the user behavior in the extracted search record is a group purchase coupon verification behavior (or other user behaviors, such as uploading UGC, connecting to WIFI), the search record can be marked as a negative sample with a smaller weight value.

[0140] Optionally, the weight values ​​of different negative samples can also be adjusted. According to different user behaviors in the feedback records, this embodiment sets different weight values ​​for different negative samples (search records in the feedback records that are not related to the target store). For example, among the negative samples, the weight value corresponding to the sample whose user behavior is group purchase coupon verification is greater than the weight value corresponding to the sample whose user behavior is click behavior.

[0141] Step S34: updating the store arrival probability prediction model according to the weighted positive samples and the weighted negative samples.

[0142] In this embodiment, all feedback records are used as samples for subsequent updating of the store arrival probability prediction model, thereby realizing the prediction process of mining feedback records from search logs and reacting to the model; weight values ​​of positive and negative samples in the feedback records are adjusted according to the determined target store that the user has arrived at and the feedback behavior generated by the user after the target store is determined, thereby ensuring the reliability of the samples used when updating the store arrival probability prediction model; when there are a large number of reliable samples, continuous iterative optimization of the store arrival probability prediction model is realized, thereby effectively improving the accuracy of the prediction results of the store arrival probability prediction model and the accuracy of the target store predicted according to the prediction results of the store arrival probability prediction model.

[0143] The store arrival prediction method in the embodiment of the present application can locate the user more accurately, improve the accuracy of distance calculation in O2O nearby search, and is applicable to a variety of scenarios where the distance from the store to the user needs to be calculated, such as the surrounding search services of various service-oriented software (including: nearby food, fresh fruits and vegetables, flower delivery, etc.), which can provide users with a better search experience. In addition, the store arrival prediction method proposed in this application is not only simple and easy to implement, and low in cost, but also can use feedback records to continuously iterate and update the store arrival prediction model, so that the accuracy of the output results of the store arrival prediction model can be higher and higher with the development of the search business scale and the accumulation of search data (for example: search records), thereby achieving more accurate positioning of users.

[0144] Based on the same inventive concept, an embodiment of the present application provides a store arrival prediction device. Figure 7 is a schematic diagram of a store arrival prediction device shown in an embodiment of the present application. Figure 7 , the device 700 comprises:

[0145] An acquisition module 701 is used to acquire a search request initiated by a user terminal;

[0146] A feature extraction module 702 is used to extract features from the search request, determine features of the user of the user terminal, features of each store in the store list corresponding to the search request, and user-store cross features, where the user-store cross features are obtained by cross-feature analysis of the features of the user and the features of each store;

[0147] The probability prediction module 703 is used to input the characteristics of the user, the characteristics of each store, and the user-store cross-characteristics into a pre-trained store arrival probability prediction model to determine the probability of the user arriving at each store;

[0148] The determination module 704 is used to determine whether the user has arrived at a target store according to the probability of the user arriving at each store, and the target store is one of the stores.

[0149] Optionally, the determining module includes:

[0150] A first determining module is used to determine that the user has arrived at the target store when the probability that the user has arrived at the target store is greater than a preset probability threshold; or

[0151] The second determination module is used to determine that the user has arrived at the target store when the probability that the user has arrived at the target store is greater than a preset probability threshold and the parameter value corresponding to the search request is within a preset effective parameter value range.

[0152] Optionally, the device further comprises:

[0153] A third determination module is used to determine the user behavior type associated with the user visiting the store;

[0154] A first extraction module is used to extract a first type of search record and a second type of search record from the search log of the user terminal, wherein the first type of search record is a search record that conforms to the user behavior type, and the second type of search record is a search record whose corresponding search time has a time difference with the search time of the first type of search record within a preset time length;

[0155] a marking module, configured to mark the search records in the first category of search records and the second category of search records that match the user behavior type as positive samples, and mark the search records in the second category of search records that do not match the user behavior type as negative samples;

[0156] The training module is used to train the preset model according to the positive samples and the negative samples to obtain the store arrival probability prediction model.

[0157] Optionally, the training module includes:

[0158] A feature extraction submodule is used to extract features from the positive samples and the negative samples respectively, and determine features of sample users corresponding to the positive samples and the negative samples, features of each sample store in the store list corresponding to the positive samples and the negative samples, and sample user-sample store cross features, where the sample user-sample store cross features are obtained by cross-feature analysis of the features of the sample users and the features of each sample store;

[0159] The training submodule is used to train the preset model using the characteristics of the sample users, the characteristics of each sample store, and the cross-features of the sample users and sample stores as training samples to obtain the store arrival probability prediction model.

[0160] Optionally, the device further comprises:

[0161] A second extraction module is used to extract, from the search log of the user terminal, the search record at the corresponding search time after it is determined that the user has arrived at the target store;

[0162] a first weight adjustment module, configured to mark the extracted search record as a positive sample and increase the weight of the positive sample when the extracted search record is a search record for the target store;

[0163] a second weight adjustment module, configured to mark the extracted search record as a negative sample and reduce the weight of the negative sample when the extracted search record is not a search record for the target store; and

[0164] The updating module is used to update the store arrival probability prediction model according to the positive samples with increased weights and the negative samples with reduced weights.

[0165] Optionally, the user's characteristics include at least one of the following: the name of the WIFI scanned or connected to by the user terminal and the corresponding signal strength, the device type of the user terminal, the longitude and latitude of the user terminal, the IP address of the user terminal, the user's user portrait, and the user's consumption preferences.

[0166] Optionally, the characteristics of each store include at least one of the following: the logo of each store, the WIFI name of each store, the average WIFI connection or scanning strength of each store, the longitude and latitude of each store, the category to which each store belongs, the price range of the goods sold in each store, the click-through rate of each store, and the visit rate of each store.

[0167] Optionally, the user-store cross feature is obtained by at least one of the following methods:

[0168] Determining the straight-line distance between the user terminal and each store according to the longitude and latitude of the user terminal and the longitude and latitude of each store;

[0169] Performing feature cross-talk on the signal strength of the WIFI of the store scanned or connected by the user terminal and the average connection or scanning strength of the WIFI of the store scanned or connected by the user terminal;

[0170] Performing feature intersection on the signal strength of the WIFI of the store scanned or connected by the user terminal and the straight-line distance between the user terminal and the store scanned or connected by the user terminal; and / or

[0171] A feature cross-talk is performed between the user click or consumption price of the user terminal and the per capita price of each store.

[0172] Based on the same inventive concept, another embodiment of the present application provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps in the method described in any of the above embodiments of the present application are implemented.

[0173] Based on the same inventive concept, another embodiment of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executed, implements the steps of the method described in any of the above embodiments of the present application.

[0174] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0175] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0176] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, devices, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0177] The embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0178] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0179] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable terminal device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0180] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present application.

[0181] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the elements.

[0182] The above is a detailed introduction to a store arrival prediction method, device, readable storage medium and electronic device provided by the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for general technical personnel in this field, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A method for predicting store arrivals, characterized in that: The method comprises: Obtaining a search request initiated by a user terminal; Extracting features from the search request to determine features of a user of the user terminal, features of each store in the store list corresponding to the search request, and user-store cross features, where the user-store cross features are obtained by cross-feature analysis of the features of the user and the features of each store; Inputting the characteristics of the user, the characteristics of each store, and the user-store cross-features into a pre-trained store arrival probability prediction model to determine the probability of the user arriving at each store; Determining whether the user has arrived at a target store according to the probability of the user arriving at each store, where the target store is one of the stores; The characteristics of each store include at least one of the following: the logo of each store, the WIFI name of each store, the average WIFI connection or scanning strength of each store, the longitude and latitude of each store, the category to which each store belongs, the price range of the goods sold by each store, the click rate of each store, and the visit rate of each store; The user-store cross-feature is obtained by at least one of the following methods: Performing feature cross-talk on the signal strength of the WIFI of the store scanned or connected by the user terminal and the average connection or scanning strength of the WIFI of the store scanned or connected by the user terminal; Performing feature intersection on the signal strength of the WIFI of the store scanned or connected by the user terminal and the straight-line distance between the user terminal and the store scanned or connected by the user terminal, wherein the straight-line distance between the store scanned or connected by the user terminal is determined according to the longitude and latitude of the user terminal and the longitude and latitude of each store; A feature cross-talk is performed between the user click or consumption price of the user terminal and the per capita price of each store.

2. The method according to claim 1, characterized in that: The step of determining whether the user has arrived at a target store according to the probability of the user arriving at each store comprises: When the probability that the user arrives at the target store is greater than a preset probability threshold, determining that the user arrives at the target store; Or when the probability that the user arrives at the target store is greater than a preset probability threshold and the parameter value corresponding to the search request is within a preset effective parameter value range, it is determined that the user arrives at the target store.

3. The method according to claim 1, characterized in that: The method further comprises: Determine the types of user behaviors associated with user visits to your store; Extracting a first category of search records and a second category of search records from the search log of the user terminal, wherein the first category of search records are search records that conform to the user behavior type, and the second category of search records are search records whose corresponding search moments have a time difference with the search moments of the first category of search records within a preset time length; Marking the search records that match the user behavior type in the first category of search records and the second category of search records as positive samples, and marking the search records that do not match the user behavior type in the second category of search records as negative samples; The preset model is trained according to the positive samples and the negative samples to obtain the store visit probability prediction model.

4. The method according to claim 3, characterized in that: The step of training a preset model according to the positive samples and the negative samples to obtain the store arrival probability prediction model comprises: Perform feature extraction on the positive samples and the negative samples respectively, determine the features of the sample users corresponding to the positive samples and the negative samples respectively, the features of each sample store in the store list corresponding to the positive samples and the negative samples respectively, and the sample user-sample store cross-features, where the sample user-sample store cross-features are obtained by cross-feature analysis of the features of the sample users and the features of each sample store; The characteristics of the sample users, the characteristics of the sample stores, and the cross-characteristics of the sample users and sample stores are used as training samples to train the preset model to obtain the store visit probability prediction model.

5. The method according to claim 2, characterized in that: After the step of determining that the user has arrived at the target store, the method further includes: Extracting, from the search log of the user terminal, a search record corresponding to the search time after it is determined that the user has arrived at the target store; In the case where the extracted search record is a search record for the target store, marking the extracted search record as a positive sample and increasing the weight of the positive sample; In the case that the extracted search record is not a search record for the target store, the extracted search record is marked as a negative sample and the weight of the negative sample is reduced; and the store arrival probability prediction model is updated based on the positive sample with increased weight and the negative sample with reduced weight.

6. The method according to claim 1, characterized in that The user characteristics include at least one of the following: the name of the WIFI scanned or connected by the user terminal and the corresponding signal strength, the device type of the user terminal, the longitude and latitude of the user terminal, the IP address of the user terminal, the user portrait of the user, and the consumption preference of the user.

7. A store arrival prediction device, characterized in that: The device comprises: An acquisition module, used to obtain a search request initiated by a user terminal; A feature extraction module, configured to extract features from the search request, determine features of the user of the user terminal, features of each store in the store list corresponding to the search request, and user-store cross features, wherein the user-store cross features are obtained by cross-feature analysis of the features of the user and the features of each store; A probability prediction module, used to input the characteristics of the user, the characteristics of each store, and the user-store cross-features into a pre-trained store arrival probability prediction model to determine the probability of the user arriving at each store; and a determination module, configured to determine whether the user has arrived at a target store according to the probability of the user arriving at each store, the target store being one of the stores; The characteristics of each store include at least one of the following: the logo of each store, the WIFI name of each store, the average WIFI connection or scanning strength of each store, the longitude and latitude of each store, the category to which each store belongs, the price range of the goods sold by each store, the click rate of each store, and the visit rate of each store; The user-store cross-feature is obtained by at least one of the following methods: Performing feature cross-talk on the signal strength of the WIFI of the store scanned or connected by the user terminal and the average connection or scanning strength of the WIFI of the store scanned or connected by the user terminal; Performing feature intersection on the signal strength of the WIFI of the store scanned or connected by the user terminal and the straight-line distance between the user terminal and the store scanned or connected by the user terminal, wherein the straight-line distance between the store scanned or connected by the user terminal is determined according to the longitude and latitude of the user terminal and the longitude and latitude of each store; A feature cross-talk is performed between the user click or consumption price of the user terminal and the per capita price of each store.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps in the method according to any one of claims 1 to 6 are implemented.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When executed, the processor implements the steps of the method according to any one of claims 1-6.