CN115456691A - Method, device, electronic device and storage medium for recommending offline advertising space - Google Patents

Abstract

The disclosure provides a method, device, electronic device, and storage medium for recommending an offline advertising space. The method includes: obtaining an advertisement placement request input by a user, and the advertisement placement request includes the type of advertisement to be placed and the POI of the advertisement to be placed. Target type; obtain user profile data and behavior track data of multiple online users; use the advertising preference estimation model to obtain multiple online users’ advertising preference for advertising types based on user profile data; analyze behavior track based on clustering algorithm Data clustering to obtain the residence labels of multiple online users; according to the advertisement preferences of multiple online users and the residence locations of the residence labels, determine the candidate geographic grid containing the residence locations Advertisement preference; determine the offline advertisement position according to the advertisement preference of the candidate geographic grid. This solution realizes the mapping between online users and offline physical space, and solves the problem of lack of effective audience data in offline advertising scenarios.

Description

Method, device, electronic device and storage medium for recommending offline advertising space

technical field

The present disclosure relates to the technical field of advertising space recommendation, and in particular to a method, device, electronic device and storage medium for recommending offline advertising space.

Background technique

In the application scenario of advertising, offline advertising is an important part. The effectiveness of advertising delivery is highly dependent on the quality of the collected audience data information. Compared with online advertising scenarios, offline advertising scenarios usually lack sufficient effective audience data information, resulting in online The recommendation accuracy of the next advertising position is not high.

Contents of the invention

In order to solve the above technical problem or at least partly solve the above technical problem, embodiments of the present disclosure provide a method, device, electronic device and storage medium for recommending an offline advertising space.

According to an aspect of the present disclosure, a method for recommending an offline advertising space is provided, including:

Acquiring an advertisement delivery request input by a user, the advertisement delivery request including the advertisement type of the advertisement to be delivered and the target type of the POI of the advertisement to be delivered;

Obtain user portrait data and behavior trajectory data of multiple online users;

According to the user portrait data, using a pre-trained advertisement preference estimation model, to obtain the advertisement preference of the plurality of online users for the advertisement type;

clustering the behavior trajectory data based on a preset clustering algorithm to obtain the residence tags of the plurality of online users;

According to the advertisement preference of the plurality of online users and the residence location corresponding to the residence label, determine the advertisement preference of the candidate geographic grid containing the residence location;

Determine a target geographic grid according to the advertisement preference of the candidate geographic grid;

A target POI of the target type contained in the target geographic grid is determined as the offline advertisement position.

According to another aspect of the present disclosure, a device for pushing offline advertising spaces is provided, including:

The first acquisition module is configured to acquire an advertisement placement request input by a user, the advertisement placement request including the advertisement type of the advertisement to be placed and the target type of the POI of the advertisement to be placed;

The second acquisition module is used to acquire user portrait data and behavior track data of multiple online users;

A third acquisition module, configured to acquire the advertisement preference of the plurality of online users for the advertisement type by using a pre-trained advertisement preference estimation model according to the user portrait data;

A clustering module, configured to cluster the behavior track data based on a preset clustering algorithm, so as to obtain the residence tags of the plurality of online users;

A first determining module, configured to determine an advertisement including a candidate geographical grid of the residence location according to the advertisement preference of the plurality of online users and the residence location corresponding to the residence label preference;

The second determining module is used to determine the target geographic grid according to the advertisement preference of the candidate geographic grid;

The third determining module is configured to determine the target POI of the target type included in the target geographic grid as the offline advertising space.

According to another aspect of the present disclosure, an electronic device is provided, including:

processor; and

memory for storing programs,

Wherein, the program includes instructions, which, when executed by the processor, cause the processor to execute the method for recommending an offline advertising space according to the aforementioned aspect.

According to another aspect of the present disclosure, there is provided a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to make the computer execute the offline advertising position according to the preceding aspect. recommended method.

According to another aspect of the present disclosure, a computer program product is provided, including a computer program, wherein when the computer program is executed by a processor, the method for recommending an offline advertising space as described in the preceding aspect is implemented.

In one or more technical solutions provided in the embodiments of the present disclosure, by obtaining the advertisement placement request input by the user, the advertisement placement request includes the advertisement type to be placed and the target type of the POI to be placed, and obtains multiple online The user's user portrait data and behavior trajectory data, and then according to the user portrait data, use the pre-trained advertisement preference prediction model to obtain the advertisement preference of multiple online users for the advertisement type, and based on the preset clustering algorithm to classify the behavior The trajectory data is clustered to obtain the residence labels of multiple online users, and then according to the advertisement preferences of multiple online users and the residence locations corresponding to the residence labels, the candidates containing the residence locations are determined The advertisement preference of the geographic grid, and then determine the target geographic grid according to the advertisement preference of the candidate geographic grid, and determine the target POI of the target type contained in the target geographic grid as an offline advertisement position. By adopting the scheme of the present disclosure, it is possible to use the user portrait data of online users to estimate the advertising preference of users in offline scenarios, and to use the behavior track data of online users to mine the residence labels of online users, and according to The permanent location determines the geographical grid of the offline physical space, and then determines the advertising preference of the geographical grid according to the advertising preference of online users, so as to determine the target POI of the target type as the recommended offline advertising position, realizing the offline The mapping between online users and offline physical space solves the problem of lack of effective audience data in offline advertising scenarios, thereby improving the accuracy of offline advertising position recommendations.

Description of drawings

Further details, features and advantages of the present disclosure are disclosed in the following description of exemplary embodiments with reference to the accompanying drawings in which:

FIG. 1 shows a flow chart of a method for recommending an offline advertising space according to an exemplary embodiment of the present disclosure;

Fig. 2 shows a flowchart of a method for recommending an offline advertising space according to another exemplary embodiment of the present disclosure;

Fig. 3 shows a schematic block diagram of an apparatus for recommending an offline advertising space according to an exemplary embodiment of the present disclosure;

FIG. 4 shows a structural block diagram of an exemplary electronic device that can be used to implement the embodiments of the present disclosure.

detailed description

Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the drawings, it should be understood that the disclosure may be embodied in various forms and should not be construed as limited to the embodiments set forth herein; A more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are for exemplary purposes only, and are not intended to limit the protection scope of the present disclosure.

It should be understood that the various steps described in the method implementations of the present disclosure may be executed in different orders, and/or executed in parallel. Additionally, method embodiments may include additional steps and/or omit performing illustrated steps. The scope of the present disclosure is not limited in this regard.

As used herein, the term "comprise" and its variations are open-ended, ie "including but not limited to". The term "based on" is "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one further embodiment"; the term "some embodiments" means "at least some embodiments." Relevant definitions of other terms will be given in the description below. It should be noted that concepts such as "first" and "second" mentioned in this disclosure are only used to distinguish different devices, modules or units, and are not used to limit the sequence of functions performed by these devices, modules or units or interdependence.

It should be noted that the modifications of "one" and "multiple" mentioned in the present disclosure are illustrative and not restrictive, and those skilled in the art should understand that unless the context clearly indicates otherwise, it should be understood as "one or more" multiple".

The names of messages or information exchanged between multiple devices in the embodiments of the present disclosure are used for illustrative purposes only, and are not used to limit the scope of these messages or information.

Before explaining the embodiments of the present disclosure, the nouns that may be involved in the present disclosure are explained as follows:

Geographic grid: a unified and simple geographical space division and positioning reference system, according to unified rules, the ground area is continuously divided according to a certain latitude and longitude or ground distance, and the spatial uncertainty is controlled within a certain range to form a regular polygon , each polygon is called a grid unit, thus forming a hierarchical and hierarchical multi-level grid system, realizing discretization of the ground space, and endowing it with a unified code;

POI: Point of Interest in geographic space, such as a community, school, shopping mall, office building, etc.;

CTR: Click-Through-Rate (Click-Through-Rate), is a commonly used term in Internet advertising, which refers to the click-through rate of online advertisements (image advertisements/text advertisements/keyword advertisements/ranking advertisements/video advertisements, etc.), that is, the advertising rate. The actual number of clicks (strictly speaking, it can be the number of reaching the target page) divided by the display volume of the advertisement (Show content).

The method, device, electronic device, and storage medium for recommending offline advertising spaces provided by the present disclosure are described below with reference to the accompanying drawings.

In the advertising delivery scenario, accurate advertising delivery is one of the core technologies. The quality of advertising reach is highly dependent on the quality of collected audience data information. However, in offline advertising scenarios, there is often a lack of sufficient effective audience data information, resulting in low accuracy of advertising.

In view of the above problems, the present disclosure provides a method for recommending offline advertising positions. By obtaining the advertisement placement request input by the user, the advertisement placement request includes the type of advertisement to be placed and the target type of the POI where the advertisement is to be placed, and obtains The user profile data and behavior trajectory data of multiple online users, and then according to the user profile data, use the pre-trained advertising preference estimation model to obtain the advertising preference of multiple online users for the type of advertising, and based on the preset aggregation The algorithm clusters the behavior trajectory data to obtain the resident labels of multiple online users, and then determines the resident location according to the advertisement preferences of multiple online users and the resident locations corresponding to the resident labels. According to the advertisement preference degree of the candidate geographical grid, the target geographical grid is determined according to the advertisement preference degree of the candidate geographical grid, and the target POI of the target type contained in the target geographical grid is determined as the offline advertisement position. By adopting the scheme of the present disclosure, it is possible to use the user portrait data of online users to estimate the advertising preference of users in offline scenarios, and to use the behavior track data of online users to mine the residence labels of online users, and according to The permanent location determines the geographical grid of the offline physical space, and then determines the advertising preference of the geographical grid according to the advertising preference of online users, so as to determine the target POI of the target type as the recommended offline advertising position, realizing the offline The mapping between online users and offline physical space solves the problem of lack of effective audience data in offline advertising scenarios, thereby improving the accuracy of offline advertising position recommendations. The disclosed solution can quickly and conveniently map online data, traffic, marketing resources and real physical space, so as to realize precise marketing location selection of any space for offline advertisements.

Fig. 1 shows a flowchart of a method for recommending an offline advertising space according to an exemplary embodiment of the present disclosure. The method can be executed by the device for recommending an offline advertising space provided in the present disclosure, wherein the device can use software and/or Hardware implementation can generally be integrated in electronic devices, including mobile phones, tablet computers, servers and other devices, and applications with online and offline advertising placement and advertising space recommendation functions can be installed in electronic devices.

As shown in Figure 1, the method for recommending an offline advertising space may include the following steps:

Step 101 , acquiring an advertisement placement request input by a user, the advertisement placement request including an advertisement type of an advertisement to be placed and a target type of a POI where the advertisement to be placed is placed.

Among them, the advertisement type to be placed may be one of multiple preset advertisement types, such as but not limited to education, beauty, clothing, shoes, bags, food, etc. Wait. The target type is one of multiple POI types, and the POI types may include, but are not limited to, residential areas, office buildings, schools, shopping malls, scenic spots, and so on. Assuming that the user wishes to place an advertisement in a community, it may be determined that the target type of the POI used for placing the advertisement is a community.

In the embodiments of the present disclosure, a user refers to an advertiser who has an advertisement placement requirement.

Exemplarily, the user can input his or her own advertisement placement requirements through the advertisement placement platform provided in the electronic device, such as the type of advertisement to be placed, where to display the advertisement, the number of desired advertisement slots, and so on. The advertisement delivery platform can generate and obtain the advertisement delivery request input by the user according to the advertisement delivery requirement input by the user, among which, according to the information input by the user where the advertisement will be displayed, the target type of the POI for placing the advertisement to be placed can be determined, for example , the user enters that the place where the ad is displayed is a school, and the target type included in the ad serving request is a school. For example, the advertising delivery platform can be an application program that supports the OMO (Online-Merge-Offline, Online-Mobile-Offline) mode, which not only supports online advertising space recommendation and advertising delivery, but also supports offline advertising space recommendation and advertising delivery.

Step 102, acquiring user portrait data and behavior track data of multiple online users.

Among them, online users can refer to the advertising audience in the online advertising scenario. For example, if a user wants to launch a newly launched course offline, he can obtain registered users on the online education platform as online users. The online education platform can use OMO mode; user portrait data can include but not limited to the user's gender, age, occupation, personal preferences, etc.; behavior track data can be obtained based on the GPS data of online users, for example, registered users of online education platforms can authorize the platform to obtain their The GPS data of the electronic devices used, the online education platform analyzes the GPS data to obtain the behavior track data of registered users.

It should be noted that in the embodiments of the present disclosure, the acquisition of the user profile data of the online users and the behavior track data of the online users is obtained under the authorization of these users.

Exemplarily, the application installed in the electronic device and configured to implement the method for recommending offline advertising spaces of the present disclosure has the characteristics of OMO business, and with the passage of time and the use of users, it has accumulated a certain amount of related business Online user data in the field, when recommending offline advertising positions, user portrait data and behavior trajectory data of multiple online users can be obtained from the accumulated online user data to describe offline advertising scenarios audience data.

Step 103, according to the user portrait data, using a pre-trained advertisement preference estimation model, to obtain the advertisement preference of the plurality of online users for the advertisement type.

Among them, the advertisement preference prediction model can be obtained by pre-training, and data such as user data in the online advertisement placement scenario, user clicks on exposed advertisements, etc. can be collected, and training samples can be constructed to train the advertisement preference prediction model.

In an optional implementation manner of the present disclosure, when the advertisement preference estimation model is obtained through training, the advertisement delivery records of the sample users in the online advertisement delivery scene and the user portrait data of the sample users may be obtained; Analyzing the advertisement delivery records to obtain historical advertisements of different advertisement types and operation data of the sample users on the historical advertisements, the operation data including clicks or no clicks; The operation data of the historically placed advertisement is marked, and the described historically placed advertisement is marked to generate a sample advertisement; the association relationship between the user portrait data of the sample user and the sample advertisements of the different advertisement types is constructed, and a training sample set is generated; An initial network model is trained based on the training sample set to obtain the advertisement preference prediction model.

Among them, the sample user can be the audience of online advertisements accumulated in the application program with OMO characteristics. For the advertisement exposed to the sample user, the sample user can choose to click on the advertisement or not to click on the advertisement according to the interest.

In the embodiment of the present disclosure, from the historical data accumulated in the online advertisement delivery scene, the advertisement delivery records of the sample users and the user portrait data corresponding to each sample user can be obtained, and the advertisement delivery records corresponding to the sample users can be obtained. Analysis to obtain the historical advertisements of different advertisement types in the advertisement records, and obtain whether the sample users have clicked on each historical advertisement, and then mark the historical advertisements according to whether the sample users click on the historical advertisements, and generate sample advertisements .

Exemplarily, when marking historical advertisements, different categories can be represented by preset identifiers, for example, the identifier "1" indicates that the sample user clicked on the advertisement, and the identifier "0" indicates that the sample user did not click on the advertisement , when generating sample advertisements, according to whether the sample user has clicked on the historical advertisements, mark "0" or "1" on the historical advertisements, and the marked advertisements are the sample advertisements.

For example, if it is determined from the advertisement delivery records corresponding to sample user A that the user has clicked on a certain historically delivered advertisement of the educational category, then the historically delivered advertisement can be marked with a mark "1" to obtain a sample advertisement .

In the embodiment of the present disclosure, for the generated sample advertisements, an association relationship between user portrait data of sample users and sample advertisements of different advertisement types may be constructed to obtain a training sample set.

Continuing with the above example, after a sample advertisement is generated, an association relationship between the user portrait data of sample user A and the sample advertisement of education can be constructed to obtain a training sample. Therefore, after constructing the association relationship between each sample user and the corresponding sample advertisement, a training sample set is obtained.

Furthermore, using the constructed training sample set, the initial network model can be trained to obtain a trained advertisement preference prediction model.

Among them, the initial network model can be a Boosting model, including but not limited to AdaBoost (adaptive boosting model), GBDT (Gradient Boosting Decision Tree, gradient boosting decision tree), XGBoost (eXtremeGradient Boosting, extreme gradient boosting tree), LightGBM (Light Gradient Boosting Machine, lightweight gradient boosting machine) and any model in CatBoost.

In the embodiment of the present disclosure, when using the training sample set to train the initial network model, the user portrait data of the sample users can be input into the initial network model as feature data, and the label of the sample advertisement can be used as the expected value, and the network model can be continuously updated The parameters are iteratively trained until the loss function value calculated according to the prediction results of the network model and the expected value is less than the preset value, the training ends, and a well-trained advertising preference prediction model is obtained. In this way, the advertising preference modeling of online users is realized, and the advertisement preference degree of different user portrait data for a certain advertisement type can be predicted by using the advertisement preference estimation model. Wherein, a higher advertisement preference degree indicates that when an advertisement of this type is served to the user, the probability that the user clicks on the advertisement is higher.

In the embodiment of the present disclosure, after obtaining the user portrait data corresponding to multiple online users, the user portrait data can be input into the trained advertisement preference estimation model, and the advertisement preference estimation model outputs the The advertisement preference degree of different advertisement types, and then obtain the advertisement preference degree of each online user for the advertisement type in the advertisement delivery request from the output result of the advertisement preference estimation model.

Step 104, clustering the behavior trajectory data based on a preset clustering algorithm to obtain the residence tags of the multiple online users.

Wherein, the clustering algorithm can be preset, and the clustering algorithm can be but not limited to DBSCAN (Density-Based Spatial Clustering of Applications with Noise, density-based spatial clustering algorithm) algorithm, agglomerative hierarchical clustering algorithm, and so on.

In the embodiment of the present disclosure, for the acquired behavior trajectory data of online users, the behavior trajectory data may be clustered based on a preset clustering algorithm, so as to obtain the residence tags of multiple online users.

Exemplarily, the resident place label can be determined according to the geographic location information corresponding to the center of the cluster obtained by clustering. For example, the place name (such as street name, county name, etc.) to which the geographic location information corresponding to the center belongs can be used as Residence label, online users belonging to a cluster have the same residence label.

Step 105, according to the advertisement preference of the plurality of online users and the residence location corresponding to the residence label, determine the advertisement preference of the candidate geographic grid including the residence location.

Wherein, the resident location corresponding to the resident label can be determined by querying the corresponding relationship between different place names and location information in the map data, and the resident location corresponding to the resident label can be expressed by latitude and longitude.

In the embodiment of the present disclosure, after determining the advertisement preference degree of each online user for the advertisement type and the residence label corresponding to each online user, it can be determined according to the inclusion relationship between the residence location and the geographic grid. The geographic grid containing the location of the resident location is used as the candidate geographic grid, and according to the resident label corresponding to each online user, the candidate geographic grid is associated with the online user, that is, the resident label and the candidate location are determined. Online users whose resident location matches the geographic grid. For example, the resident label corresponding to online user B is C, and the resident location contained in the candidate geographic grid a corresponds to C, then online user B Associated with the candidate geographic grid a, the online user B is a user resident in the candidate geographic grid. Thus, the spatial aggregation of different users is realized. Furthermore, for each candidate geographic grid, the advertisement preference of the candidate geographic grid may be determined according to the advertisement preference of all online users associated with the candidate geographic grid.

Exemplarily, when determining the advertisement preference of each candidate geographic grid, for a candidate geographic grid, the largest advertisement preference can be selected from the advertisement preference of all online users associated with the candidate geographic grid as the advertisement preference of the candidate geographic grid.

Step 106: Determine a target geographic grid according to the advertisement preference of the candidate geographic grid.

In the embodiment of the present disclosure, after the advertisement preference of each candidate geographic grid is determined, the target geographic grid may be determined from the candidate geographic grids according to the advertisement preference of each candidate geographic grid.

Exemplarily, after the advertisement preference degree of each candidate geographic grid is determined, each candidate geographic grid can be sorted according to the order of advertisement preference from large to small, and obtain the top preset number of Candidate geographic grids serve as target geographic grids. Wherein, the preset number can be preset according to actual needs, or the preset number can be determined according to the advertising space budget in the advertisement delivery request input by the user. When the user does not input the advertising space budget, the preset The default value is used as the preset number.

Exemplarily, after the advertisement preference degree of each candidate geographic grid is determined, a preference average value is calculated according to the advertisement preference degrees of all candidate geographic grids, and the advertisement preference degree in the candidate geographic grid is greater than the preference Candidate geographic grids with degree mean values are determined as target geographic grids.

Step 107, determining the target POI of the target type contained in the target geographic grid as the offline advertising space.

In the embodiment of the present disclosure, after the target geographic grid is determined, the target POI of the target type can be obtained from the target geographic grid, and the acquired target POI can be determined as an offline advertisement position, and the determined offline advertisement can be recommended to the user .

Exemplarily, when recommending a certain offline advertisement position to the user, information such as the name of the target POI and the geographical location may be displayed.

The method for recommending an offline advertising space in an embodiment of the present disclosure obtains the advertisement placement request input by the user, the advertisement placement request includes the advertisement type to be placed and the target type of the POI to place the advertisement, and obtains multiple online users. According to the user portrait data and behavior trajectory data, and then according to the user portrait data, use the pre-trained advertising preference estimation model to obtain the advertising preference of multiple online users for the type of advertisement, and based on the preset clustering algorithm to analyze the behavior trajectory The data is clustered to obtain the resident labels of multiple online users, and then according to the advertisement preferences of multiple online users and the resident locations corresponding to the resident labels, determine the candidate geographic location that includes the resident location The advertisement preference of the grid, and according to the advertisement preference of the candidate geographic grid, determine the target geographic grid, and determine the target POI of the target type contained in the target geographic grid as the offline advertisement position. By adopting the scheme of the present disclosure, it is possible to use the user portrait data of online users to estimate the advertising preference of users in offline scenarios, and to use the behavior track data of online users to mine the residence labels of online users, and according to The permanent location determines the geographical grid of the offline physical space, and then determines the advertising preference of the geographical grid according to the advertising preference of online users, so as to determine the target POI of the target type as the recommended offline advertising position, realizing the offline The mapping between online users and offline physical space solves the problem of lack of effective audience data in offline advertising scenarios, thereby improving the accuracy of offline advertising position recommendations.

In an optional implementation of the present disclosure, according to the advertisement preference of multiple online users and the residence location corresponding to the residence label, determine the advertisement preference of the candidate geographic grid containing the residence location , according to the resident location corresponding to the resident label, the candidate geographic grid containing the resident location can be determined from multiple geographic grids; then, for each candidate geographic grid, the candidate geographic grid The advertisement preference of online users contained in the grid is accumulated to obtain the advertisement preference of each candidate geographic grid.

Exemplarily, the plurality of geographic grids may be a plurality of grid units obtained by dividing a real geographic area based on a preset geographic spatial point indexing algorithm.

Exemplarily, the plurality of geographic grid units may be obtained by screening the plurality of grid units obtained by dividing the real geographical area based on the preset geographic spatial point indexing algorithm according to a preset screening strategy.

In the embodiment of the present disclosure, after determining the advertisement preference of each online user for the type of advertisement and the residence label corresponding to each online user, the residence location corresponding to the residence label can be firstly determined, Determine the candidate geographic grid containing the resident location from multiple geographic grids, and then determine the All online users contained in each candidate geographic grid, and then for each candidate geographic grid, the advertisement preference of all online users contained in the same candidate geographic grid is accumulated to obtain the candidate geographic grid Advertising preference, so as to obtain the advertising preference of each candidate geographic grid.

Exemplarily, assuming that a candidate geographic grid includes 30 online users, and the advertisement preference of these 30 online users is all 0.4, the advertisement preference of the candidate geographic grid is 30*0.4=12.

In the embodiment of the present disclosure, according to the resident location corresponding to the resident label, the candidate geographic grid containing the resident location is determined from multiple geographic grids, and then for each candidate geographic grid, the The advertising preference of online users contained in the candidate geographic grids is accumulated to obtain the advertising preference of each candidate geographic grid. Therefore, according to the spatial inclusion relationship between the resident place and the geographic grid, the geographic network-based Based on the estimation of crowd advertising preference in the grid, the advertising preference corresponding to the geographic grid is determined, and the mapping between online data and offline physical space is realized, which provides data support for the precise selection of offline advertising positions.

In an optional implementation manner of the present disclosure, the multiple geographic grids described in the foregoing embodiments may be determined through an implementation manner as shown in FIG. 2 . Fig. 2 shows a flow chart of a method for recommending an offline advertising space according to another exemplary embodiment of the present disclosure. As shown in Fig. 2 , on the basis of the foregoing embodiments, the method for recommending an offline advertising space may also be Include the following steps:

Step 201 , acquiring a plurality of grid units with a preset size and containing POIs of the target type.

Wherein, the preset size may be the side length, radius, etc. of the grid unit, or may be the area of the grid unit, which is not limited in the present disclosure.

Exemplarily, assuming that the preset size is the side length, the position of the POI of the target type in the delivery area (such as one or more cities) expected by the user can be obtained, and the position of the POI of the target type is the center, with The preset size is used as the side length, and a square area with a side length of the preset size is divided around the POI as a grid unit, and multiple POIs can obtain multiple grid units.

For example, assuming that the preset size is the area, the position of the target type POI in the user's expected delivery area can be obtained, and with the position of the target type POI as the center, an area is divided around the POI as the preset size The region of is used as a grid unit, and multiple POIs get multiple grid units.

In an optional implementation of the present disclosure, when obtaining multiple grid units, the geospatial area can also be divided into multiple grid areas based on the geospatial point index algorithm first, and the grid area and its grid POI association in the area; then filter out the candidate grid area containing the POI of the target type from multiple grid areas; if the size of the candidate grid area is less than or equal to the preset size, determine the candidate grid area as a grid unit ; If the size of the candidate grid area is greater than the preset size, the candidate grid area is divided to obtain multiple sub-grids; for each sub-grid, it can be judged whether the size of the sub-grid is greater than the preset size, if the The size of the sub-grid is not greater than the preset size and contains the POI of the target type, then determining the sub-grid as the grid unit; if the size of the sub-grid is greater than the preset size , then the candidate sub-grids containing the POI of the target type will continue to be divided until a grid unit not larger than the preset size and containing the POI of the target type is obtained.

Among them, the geospatial point index algorithm can be but not limited to Genhash algorithm, Google S2 algorithm, etc.; the geospatial area can be the geographic space of the whole country, or it can be the expected area specified by the user (advertiser) who wants to place advertisements. There is no limit to this publicly.

Exemplarily, Google S2 algorithm can be used to maintain a set of hierarchical geographic grid ID system for geospatial areas, including multiple grid areas and corresponding IDs, and obtain POIs (including POI names, type, location information, etc.), and associate the POI with the ID of the grid area to which it belongs.

It can be understood that among the multiple grid areas divided based on the geospatial point index algorithm, some grids may not contain the POI of the target type, so in the embodiment of the present disclosure, these grid areas can be filtered out, Only the grid area containing the POI of the target type among the grid areas is reserved as the candidate grid area, thereby realizing the preliminary screening of the grid area.

Next, for the screened candidate grid area, it may be determined whether the size of the candidate grid area satisfies (is not greater than) a preset size. For example, assuming that the preset size is the side length, you can compare whether the maximum side length of the candidate grid area is less than or equal to the preset side length threshold, and if so, consider that the side length of the candidate grid area is not greater than the preset side length threshold , otherwise it is considered that the side length of the candidate grid region is greater than a preset side length threshold, wherein the preset side length threshold can be preset according to actual needs. For another example, assuming that the preset size is an area, it is possible to compare whether the area of the candidate grid region is greater than the preset area threshold, if so, the area of the candidate grid region is greater than the preset area threshold, otherwise the area of the candidate grid region is considered to be larger than the preset area threshold. greater than a preset area threshold, wherein the preset area threshold can be preset according to actual needs.

In the embodiment of the present disclosure, if the size of the candidate grid area is smaller than or equal to the preset size, the candidate grid area is determined as a grid unit; if the size of the candidate grid area is larger than the preset size, the candidate grid area can be The grid area is divided to obtain multiple sub-grids. For example, the candidate grid area can be quartered to obtain four sub-grids. For each divided sub-grid, it may be determined whether the size of the sub-grid is larger than a preset size, and the judgment method is similar to the method of judging whether the size of the candidate grid area is larger than the preset size. If the size of the sub-grid is less than or equal to the preset size, and the sub-grid contains POIs of the target type, the sub-grid is determined as a grid unit; if the sub-grid does not contain POIs of the target type, it is filtered out The sub-grid; if the sub-grid contains POIs of the target type, but the size of the sub-grid is larger than the preset size, continue to divide the sub-grid to obtain multiple smaller sub-grids, and repeat the above process , until a grid unit not larger than the preset size and containing the POI of the target type is obtained.

In an optional implementation of the present disclosure, for any candidate grid area that contains the POI of the target type, its area may be compared with a preset area threshold, and if its area is less than or equal to the preset area threshold , the candidate grid area is determined as a grid unit; if its area is greater than the preset area threshold, the candidate grid area is divided to obtain multiple sub-grids. Afterwards, the sub-grids that do not contain the POI of the target type in the multiple sub-grids are discarded, and the area of the sub-grids that contain the POIs of the target type is compared with the preset area threshold, if the first sub-grid in the multiple sub-grids When the area of the grid is not greater than the preset area threshold and contains the POI of the target type, the first sub-grid is determined as the grid unit, wherein the first sub-grid is any sub-grid in the plurality of sub-grids ; If the area of the second sub-grid in the multiple sub-grids is greater than the preset area threshold and contains the POI of the target type, continue to divide the second sub-grid to obtain multiple smaller sub-grids, and repeat the above process , until a grid cell whose area is not greater than the preset area threshold and contains the POI of the target type is obtained, wherein the second sub-grid is any sub-grid in the plurality of sub-grids except the first sub-grid.

That is to say, among the multiple sub-grids obtained by segmenting the candidate grid area, for the first sub-grid whose area is not greater than the preset area threshold and contains POIs of the target type, the first sub-grid is directly determined as the grid grid unit; for the second subgrid whose area is greater than the preset area threshold and contains the POI of the target type, it is further divided into smaller multiple subgrids, and the above process is repeated until the area obtained is less than the preset area threshold and Grid cell containing POIs of the target type.

In the embodiment of the present disclosure, by constructing the association relationship between the grid area and the POI, and scanning in the geospatial area to filter the grid unit of the preset size and containing the POI of the target type, the main solution of point resources and traffic coupling, and grid search can be accelerated through grid division.

Step 202: Determine the score of each grid unit according to the number of POIs of each POI type contained in each grid unit.

In an optional implementation manner of the present disclosure, for each grid unit, the POI types contained in the grid unit and the number of POIs of each POI type can be counted, and the number of POIs of different POI types is accumulated The value serves as the score for that grid cell. For example, if a grid unit contains 2 schools and 1 shopping mall, then the total number of POIs (2+1=3) included will be used as the score of the grid unit.

In an optional implementation of the present disclosure, for each grid unit, the number of POIs of each POI type contained in the grid unit can be obtained; then, according to the preset scores corresponding to different POI types and The number of POIs of each POI type included in the grid unit determines the score of each grid unit. For example, the preset scores corresponding to different POI types are: 1 point for schools, 0.8 points for residences, 0.6 points for office buildings, and 0.5 points for shopping malls. Assuming that a grid unit includes 2 schools and 1 shopping mall, the grid unit The corresponding score is: 1*2+0.5*1=2.5 points.

Step 203: Determine the plurality of geographic grids from the plurality of grid units according to the score of each grid unit.

In the embodiment of the present disclosure, after the score corresponding to each grid unit is determined, multiple geographic grids may be determined from multiple grid units according to the score corresponding to each grid unit.

In an optional implementation of the present disclosure, a preset value can be set in advance as the number of geographic grids to be acquired, then after determining the score of each grid unit, it can be calculated according to the score from high to low In order, the grid units with preset values are selected from the grid units as the determined plurality of geographic grids, that is, the grid units with the highest scores and preset values are selected as the geographic grids.

In an optional implementation of the present disclosure, the advertisement placement request input by the user may include the number of budget advertisements, which is the number of advertisements that the user expects to be recommended. The budget slots determine grid cells that contain a sufficient number of POIs of the target type as multiple geographic grids. Therefore, in the embodiment of the present disclosure, when multiple geographic grids are determined from multiple grid units, the candidate advertisement digits can be determined first according to the budget advertisement digits and the preset multiple; Sorting the multiple grid units in descending order, and starting from the grid unit sorted first, sequentially accumulating the total number of POIs of the target type contained in each grid unit until all When the total number reaches the number of candidate advertisement positions, the currently accumulated target grid unit is determined as the plurality of geographical grids.

Wherein, the preset multiple can be preset according to actual needs, for example, the preset multiple can be set to 6 times, 10 times, etc.

For example, assuming that the number of budget advertisements entered by the user is 20 and the preset multiplier is 5, the number of candidate advertisements is determined to be 100, that is, the number of POIs of the target type contained in multiple geographic grids should be no less than 100 indivual. In the embodiment of the present disclosure, after the score of each grid unit is determined, the grid units are sorted according to the order of the score from high to low, and starting from the grid unit sorted first, the grid units are accumulated sequentially The total number of POIs of the target type contained in the unit and subsequent grid units. When the total number reaches 100, the current accumulated target grid unit is determined to be multiple geographic grids, that is, it will contain the first The grid cells of 100 POIs of the target type and the grid cells before them are determined as a plurality of geographic grids. For example, assuming that the number of POIs of the target type contained in the first grid unit after sorting is added, and when added to the 35th grid unit of the sort, the total number of POIs of the target type reaches 100, then the sorting is determined The 35 grid cells from the 1st to the 35th are geographic grids. Thus, grid units with higher scores and containing POIs of the target type can be determined as a plurality of geographic grids, and it can be ensured that the finally recommended advertising positions meet the quantity requirements of users.

In addition, in an optional implementation of the present disclosure, if the cumulative total number of POIs of the target type contained in all grid units is still less than the number of candidate advertisements, then all grid units are determined as geographical grid units. grid.

The method for recommending an offline advertising space in an embodiment of the present disclosure obtains multiple grid units of a preset size and contains POIs of the target type, and determines each POI according to the number of POIs of each POI type contained in each grid unit The score of the grid unit, and then according to the score of each grid unit, determine multiple geographic grids from multiple grid units, thus realizing scoring for each grid unit, and according to the score from the grid The geographical grid is screened out in the unit for subsequent association with the residence label of the online user, which can reduce the number of geographical grids that need to be compared with the residence location, thereby improving the comparison speed and efficiency.

By adopting the disclosed scheme, online data, traffic, marketing resources, etc. can be associated and mapped with offline real physical space quickly and conveniently, so as to realize accurate marketing position selection of any spatial position of offline advertisements.

The exemplary embodiment of the present disclosure also provides an apparatus for recommending an offline advertising space. Fig. 3 shows a schematic block diagram of an apparatus for recommending an offline advertising space according to an exemplary embodiment of the present disclosure. As shown in Fig. 3 , the apparatus 30 for recommending an offline advertising space includes: a first acquisition module 310, a An acquisition module 320 , a third acquisition module 330 , a clustering module 340 , a first determination module 350 , a second determination module 360 and a third determination module 370 .

Wherein, the first acquiring module 310 is configured to acquire an advertisement placement request input by a user, the advertisement placement request including the advertisement type of the advertisement to be placed and the target type of the POI of the advertisement to be placed;

The second obtaining module 320 is used to obtain user portrait data and behavior track data of multiple online users;

The third acquisition module 330 is configured to acquire the advertisement preference of the plurality of online users for the advertisement type by using a pre-trained advertisement preference estimation model according to the user portrait data;

A clustering module 340, configured to cluster the behavior trajectory data based on a preset clustering algorithm, so as to obtain the residence tags of the plurality of online users;

The first determination module 350 is configured to determine, according to the advertisement preferences of the multiple online users and the residence location corresponding to the residence label, the location of the candidate geographic grid containing the residence location Ad preference;

The second determining module 360 is configured to determine a target geographic grid according to the advertisement preference of the candidate geographic grid;

The third determining module 370 is configured to determine the target POI of the target type contained in the target geographic grid as the offline advertising position.

Optionally, the first determining module 350 is further configured to:

Determine a candidate geographic grid containing the resident location from a plurality of geographic grids according to the resident location corresponding to the resident label;

For each of the candidate geographic grids, the advertisement preference degrees of the online users included in the candidate geographic grids are accumulated to obtain the advertisement preference degrees of each of the candidate geographic grids.

Optionally, the device 30 for recommending offline advertising spaces further includes:

A fourth acquisition module, configured to acquire a plurality of grid units of a preset size and containing POIs of the target type;

A fourth determination module, configured to determine the score of each grid unit according to the number of POIs of each POI type contained in each grid unit;

The fifth determination module is configured to determine the plurality of geographic grids from the plurality of grid units according to the score of each grid unit.

Optionally, the fourth acquisition module is also used for:

dividing the geospatial area into a plurality of grid areas based on a geospatial point indexing algorithm, the grid areas being associated with POIs located within their grid areas;

selecting a candidate grid area containing POIs of the target type from the plurality of grid areas;

In the case that the area of the candidate grid area is greater than a preset size area threshold, the candidate grid area is divided to obtain a plurality of sub-grids;

If the area of the first sub-grid among the plurality of sub-grids is not greater than the preset area threshold and contains the POI of the target type, determine the first sub-grid as the grid unit , wherein the first subgrid is any subgrid in the plurality of subgrids;

In the case where the area of the second sub-grid among the plurality of sub-grids is greater than the preset area threshold and contains POIs of the target type, continue to divide the second sub-grid until no larger than The preset area threshold and the grid unit containing the POI of the target type, wherein the second sub-grid is any sub-grid in the plurality of sub-grids except the first sub-grid .

Optionally, the fourth determining module is also used for:

For each grid unit, obtain the number of POIs of each POI type contained in the grid unit;

The score of each grid unit is determined according to the preset scores corresponding to different POI types and the number of POIs of each POI type contained in the grid unit.

Optionally, the advertisement placement request also includes a budgeted number of advertisements; the fifth determination module is also used for:

Determine the number of candidate advertisements according to the number of advertisements in the budget and the preset multiple;

The plurality of grid units are sorted in descending order of scores, and starting from the first sorted grid unit, the total number of POIs of the target type contained in each grid unit is sequentially accumulated. until the total number reaches the number of candidate advertisement positions, and determine the currently accumulated target grid unit as the plurality of geographic grids.

Optionally, the device 30 for recommending offline advertising spaces further includes: a training module; the training module is used for:

Obtaining the advertisement delivery records of the sample users in the online advertisement delivery scenario, and the user portrait data of the sample users;

Analyzing the advertisement delivery records to obtain historical advertisements of different advertisement types and operation data of the sample users on the historical advertisements, the operation data including clicks or no clicks;

Marking the historical advertisements according to the operation data of the sample users on the historical advertisements to generate sample advertisements;

Constructing the association relationship between the user portrait data of the sample user and the sample advertisements of different advertisement types, and generating a training sample set;

An initial network model is trained based on the training sample set to obtain the advertisement preference prediction model.

The device for recommending offline advertising spaces provided by the embodiments of the present disclosure can execute any method for recommending offline advertising spaces applicable to electronic devices provided by the embodiments of the present disclosure, and has corresponding functional modules and beneficial effects for executing the methods. For the content not described in detail in the device embodiment of the present disclosure, reference may be made to the description in any method embodiment of the present disclosure.

Exemplary embodiments of the present disclosure also provide an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor. The memory stores a computer program capable of being executed by the at least one processor, and when the computer program is executed by the at least one processor, the electronic device executes an offline advertising space according to an embodiment of the present disclosure. recommended method.

Exemplary embodiments of the present disclosure also provide a non-transitory computer-readable storage medium storing a computer program, wherein, when the computer program is executed by a processor of a computer, the computer is used to cause the computer to execute the Recommended method for offline advertising slots.

Exemplary embodiments of the present disclosure also provide a computer program product, including a computer program, wherein, when the computer program is executed by a processor of a computer, the computer is used to cause the computer to execute the offline advertising position according to the embodiment of the present disclosure. recommended method.

Referring to FIG. 4 , a structural block diagram of an electronic device 1100 that can serve as a server or a client of the present disclosure will now be described, which is an example of a hardware device that can be applied to various aspects of the present disclosure. Electronic device is intended to mean various forms of digital electronic computing equipment, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other suitable computers. Electronic devices may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are by way of example only, and are not intended to limit implementations of the disclosure described and/or claimed herein.

As shown in FIG. 4 , an electronic device 1100 includes a computing unit 1101, which can perform calculations according to a computer program stored in a read-only memory (ROM) 1102 or a computer program loaded from a storage unit 1108 into a random access memory (RAM) 1103. Various appropriate actions and processes are performed. In the RAM 1103, various programs and data necessary for the operation of the device 1100 can also be stored. The computing unit 1101 , ROM 1102 , and RAM 1103 are connected to each other through a bus 1104 . An input/output (I/O) interface 1105 is also connected to the bus 1104 .

Multiple components in the electronic device 1100 are connected to the I/O interface 1105 , including: an input unit 1106 , an output unit 1107 , a storage unit 1108 and a communication unit 1109 . The input unit 1106 can be any type of device capable of inputting information to the electronic device 1100. The input unit 1106 can receive input digital or character information, and generate key signal input related to user settings and/or function control of the electronic device. The output unit 1107 may be any type of device capable of presenting information, and may include, but is not limited to, a display, a speaker, a video/audio output terminal, a vibrator, and/or a printer. The storage unit 1108 may include, but is not limited to, a magnetic disk and an optical disk. The communication unit 1109 allows the electronic device 1100 to exchange information/data with other devices through a computer network such as the Internet and/or various telecommunication networks, and may include but not limited to a modem, a network card, an infrared communication device, a wireless communication transceiver and/or a chip Groups, such as Bluetooth™ devices, WiFi devices, WiMax devices, cellular communication devices, and/or the like.

The computing unit 1101 may be various general-purpose and/or special-purpose processing components having processing and computing capabilities. Some examples of computing units 1101 include, but are not limited to, central processing units (CPUs), graphics processing units (GPUs), various dedicated artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, digital signal processing processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 1101 executes the various methods and processes described above. For example, in some embodiments, the method for recommending offline advertising spots may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as storage unit 1108 . In some embodiments, part or all of the computer program may be loaded and/or installed on the electronic device 1100 via the ROM 1102 and/or the communication unit 1109 . In some embodiments, the computing unit 1101 may be configured in any other appropriate way (for example, by means of firmware) to execute a method for recommending an offline advertising space.

Program codes for implementing the methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special purpose computer, or other programmable data processing devices, so that the program codes, when executed by the processor or controller, make the functions/functions specified in the flow diagrams and/or block diagrams Action is implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package partly on the machine and partly on a remote machine or entirely on the remote machine or server.

In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media would include one or more wire-based electrical connections, portable computer discs, hard drives, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM or flash memory), optical fiber, compact disk read only memory (CD-ROM), optical storage, magnetic storage, or any suitable combination of the foregoing.

As used in this disclosure, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, apparatus, and/or means for providing machine instructions and/or data to a programmable processor (eg, magnetic disk, optical disk, memory, programmable logic device (PLD)), including machine-readable media that receive machine instructions as machine-readable signals. The term "machine-readable signal" refers to any signal used to provide machine instructions and/or data to a programmable processor.

To provide for interaction with the user, the systems and techniques described herein can be implemented on a computer having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user. ); and a keyboard and pointing device (eg, a mouse or a trackball) through which a user can provide input to the computer. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and can be in any form (including Acoustic input, speech input or, tactile input) to receive input from the user.

The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., as a a user computer having a graphical user interface or web browser through which a user can interact with embodiments of the systems and techniques described herein), or including such backend components, middleware components, Or any combination of front-end components in a computing system. The components of the system can be interconnected by any form or medium of digital data communication, eg, a communication network. Examples of communication networks include: Local Area Network (LAN), Wide Area Network (WAN) and the Internet.

A computer system may include clients and servers. Clients and servers are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by computer programs running on the respective computers and having a client-server relationship to each other.

Claims (10)

1. A method for recommending an offline advertising space, wherein the method comprises: Acquiring an advertisement delivery request input by a user, the advertisement delivery request including the advertisement type of the advertisement to be delivered and the target type of the POI of the advertisement to be delivered; Obtain user portrait data and behavior trajectory data of multiple online users; According to the user portrait data, using a pre-trained advertisement preference estimation model, to obtain the advertisement preference of the plurality of online users for the advertisement type; clustering the behavior trajectory data based on a preset clustering algorithm to obtain the residence tags of the plurality of online users; According to the advertisement preference of the plurality of online users and the residence location corresponding to the residence label, determine the advertisement preference of the candidate geographic grid containing the residence location; Determine a target geographic grid according to the advertisement preference of the candidate geographic grid; A target POI of the target type contained in the target geographic grid is determined as the offline advertisement position. 2. The method for recommending an offline advertising space according to claim 1, wherein, according to the advertisement preference degree of the plurality of online users and the resident location corresponding to the resident label, determine Advertisement preferences of candidate geographic grids containing the resident location, including: Determine a candidate geographic grid containing the resident location from a plurality of geographic grids according to the resident location corresponding to the resident label; For each of the candidate geographic grids, the advertisement preference degrees of the online users included in the candidate geographic grids are accumulated to obtain the advertisement preference degrees of each of the candidate geographic grids. 3. The method for recommending an offline advertising space as claimed in claim 2, wherein said method further comprises: Obtaining a plurality of grid units of a preset size and containing POIs of the target type; determining the score of each grid unit according to the number of POIs of each POI type contained in each grid unit; The plurality of geographic grids are determined from the plurality of grid units based on the score of each of the grid units. 4. The method for recommending an offline advertising space according to claim 3, wherein said obtaining a plurality of grid units of a preset size and containing POIs of said target type comprises: dividing the geospatial area into a plurality of grid areas based on a geospatial point indexing algorithm, the grid areas being associated with POIs located within their grid areas; selecting a candidate grid area containing POIs of the target type from the plurality of grid areas; When the area of the candidate grid area is greater than a preset area threshold, segment the candidate grid area to obtain a plurality of sub-grids; If the area of the first sub-grid among the plurality of sub-grids is not greater than the preset area threshold and contains the POI of the target type, determine the first sub-grid as the grid unit , wherein the first subgrid is any subgrid in the plurality of subgrids; In the case where the area of the second sub-grid among the plurality of sub-grids is greater than the preset area threshold and contains POIs of the target type, continue to divide the second sub-grid until no larger than The preset area threshold and the grid unit containing the POI of the target type, wherein the second sub-grid is any sub-grid in the plurality of sub-grids except the first sub-grid . 5. The method for recommending an offline advertising space according to claim 3, wherein, according to the number of POIs of each POI type contained in each grid unit, the score of each grid unit is determined, include: For each grid unit, obtain the number of POIs of each POI type contained in the grid unit; The score of each grid unit is determined according to the preset scores corresponding to different POI types and the number of POIs of each POI type contained in the grid unit. 6. The method for recommending an offline advertising space according to claim 3, wherein the advertisement delivery request further includes a budgeted number of advertisements; And wherein, according to the score of each of the grid units, determining a plurality of geographic grids from the plurality of grid units includes: Determine the number of candidate advertisements according to the number of advertisements in the budget and the preset multiple; The plurality of grid units are sorted in descending order of scores, and starting from the first sorted grid unit, the total number of POIs of the target type contained in each grid unit is sequentially accumulated. until the total number reaches the number of candidate advertisement positions, and determine the currently accumulated target grid unit as the plurality of geographic grids. 7. The method for recommending an offline advertising space according to any one of claims 1-6, wherein the advertisement preference prediction model is obtained through training in the following manner: Obtaining the advertisement delivery records of the sample users in the online advertisement delivery scenario, and the user portrait data of the sample users; Analyzing the advertisement delivery records to obtain historical advertisements of different advertisement types and operation data of the sample users on the historical advertisements, the operation data including clicks or no clicks; Marking the historical advertisements according to the operation data of the sample users on the historical advertisements to generate sample advertisements; Constructing the association relationship between the user portrait data of the sample user and the sample advertisements of different advertisement types, and generating a training sample set; An initial network model is trained based on the training sample set to obtain the advertisement preference prediction model. 8. A device for recommending an offline advertising space, wherein the device comprises: The first acquisition module is configured to acquire an advertisement placement request input by a user, the advertisement placement request including the advertisement type of the advertisement to be placed and the target type of the POI of the advertisement to be placed; The second acquisition module is used to acquire user portrait data and behavior track data of multiple online users; A third acquisition module, configured to acquire the advertisement preference of the plurality of online users for the advertisement type by using a pre-trained advertisement preference estimation model according to the user portrait data; A clustering module, configured to cluster the behavior track data based on a preset clustering algorithm, so as to obtain the residence tags of the plurality of online users; A first determining module, configured to determine an advertisement including a candidate geographical grid of the residence location according to the advertisement preference of the plurality of online users and the residence location corresponding to the residence label preference; The second determining module is used to determine the target geographic grid according to the advertisement preference of the candidate geographic grid; The third determining module is configured to determine the target POI of the target type included in the target geographic grid as the offline advertising space. 9. An electronic device comprising: processor; and memory for storing programs, Wherein, the program includes instructions, which, when executed by the processor, cause the processor to execute the method for recommending an offline advertising space according to any one of claims 1-7. 10. A non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to make the computer execute the method for recommending offline advertising positions according to any one of claims 1-7 .

Images