Address recommendation method and related equipment

Abstract

The invention provides an address recommendation method and related equipment. The method comprises the steps of obtaining a target position; determining candidate addresses in the addresses based on the target position and address popularity information of the addresses in an address database; wherein the address popularity information is used for indicating the attention degree of the address; sorting and scoring the candidate addresses based on a preset dimension to obtain a sorting score; and determining a preset number of candidate addresses with the maximum sorting score as target addresses corresponding to the target position.

Description

Technical Field

[0001] The present disclosure relates to the field of data processing, and in particular to an address recommendation method and related equipment. Background Art

[0002] In scenarios where addresses are needed, recommended addresses can be provided based on location or partial address information entered, with authorization, so that address users can directly select them, simplifying operations such as manual address entry. However, recommended addresses are often inaccurate, resulting in a low adoption rate of recommended addresses. Summary of the invention

[0003] The present disclosure proposes an address recommendation method and related devices, which at least to a certain extent solve the technical problems of low accuracy and low adoption rate of recommended addresses in the related art.

[0004] In a first aspect, the present disclosure provides an address recommendation method, comprising:

[0005] Get the target location;

[0006] Determining a candidate address from the addresses based on the target location and address heat information of addresses in an address database; wherein the address heat information is used to indicate the attention level of the address;

[0007] Sorting and scoring the candidate addresses based on a preset dimension to obtain a ranking score;

[0008] A preset number of the candidate addresses with the largest ranking scores are determined as target addresses corresponding to the target positions, and the target addresses are output.

[0009] In a second aspect of the present disclosure, there is provided an address recommendation device, comprising:

[0010] An acquisition module is used to obtain a target position;

[0011] A recall module, configured to determine a candidate address from the addresses based on the target location and address heat information of the addresses in the address database; wherein the address heat information is used to indicate the attention level of the address;

[0012] A sorting module, used to sort and score the candidate addresses based on a preset dimension to obtain a sorting score;

[0013] The recommendation module is used to determine a preset number of the candidate addresses with the largest ranking scores as target addresses corresponding to the target position, and output the target addresses.

[0014] According to a third aspect of the present disclosure, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method described in the first aspect is implemented.

[0015] According to a fourth aspect of the present disclosure, a non-transitory computer-readable storage medium is provided, wherein the non-transitory computer-readable storage medium stores computer instructions, wherein the computer instructions are used to cause the computer to execute the method described in the first aspect.

[0016] According to a fifth aspect of the present disclosure, a computer program product is provided, comprising computer program instructions, which, when executed on a computer, cause the computer to execute the method according to the first aspect.

[0017] As can be seen from the above, the address recommendation method and related devices provided by the present disclosure obtain the target location and combine the address popularity information in the address database to perform preliminary screening to obtain candidate addresses. The candidate addresses are then sorted to determine the target address corresponding to the target location, which can improve the accuracy of address recommendation and the adoption rate of recommended addresses. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the present disclosure or related technologies, the drawings required for use in the embodiments or related technical descriptions are briefly introduced below. Obviously, the drawings described below are only embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0019] Figure 1 A schematic diagram of an address recommendation architecture according to an embodiment of the present disclosure.

[0020] Figure 2 The figure is a schematic diagram of the hardware structure of an exemplary electronic device according to an embodiment of the present disclosure.

[0021] Figure 3 The figure is a flowchart of the address recommendation method according to an embodiment of the present disclosure.

[0022] Figure 4 A schematic diagram of an address recommendation method according to an embodiment of the present disclosure.

[0023] Figure 5 A schematic diagram of a sorting model according to an embodiment of the present disclosure.

[0024] Figure 6 Schematic diagram of an address recommendation device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0025] In order to make the objectives, technical solutions and advantages of the present disclosure more clearly understood, the present disclosure is further described in detail below in combination with specific embodiments and with reference to the accompanying drawings.

[0026] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present disclosure should be understood by people with ordinary skills in the field to which the present disclosure belongs. The "first", "second" and similar words used in the embodiments of the present disclosure do not indicate any order, quantity or importance, but are only used to distinguish different components. "Including" or "comprising" and similar words mean that the elements or objects appearing before the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0027] It is understandable that before using the technical solutions disclosed in the embodiments of the present disclosure, the type, scope of use, and usage scenarios of the personal information involved in the present disclosure should be informed to the user and the user's authorization should be obtained in an appropriate manner in accordance with relevant laws and regulations. For example, in response to receiving an active request from the user, a prompt message is sent to the user to clearly prompt the user that the operation requested to be performed will require the acquisition and use of the user's personal information. Thus, the user can autonomously choose whether to provide personal information to software or hardware such as an electronic device, application, server, or storage medium that performs the operation of the technical solution of the present disclosure according to the prompt message.

[0028] It is understandable that the above notification and the process of obtaining user authorization are merely illustrative and do not constitute a limitation on the implementation of the present disclosure. Other methods that meet the relevant laws and regulations may also be applied to the implementation of the present disclosure.

[0029] Figure 1 A schematic diagram of an address recommendation architecture according to an embodiment of the present disclosure is shown. Figure 1 , the address recommendation architecture 100 may include a server 110, a terminal 120, and a network 130 providing a communication link. The server 110 and the terminal 120 may be connected via a wired or wireless network 130. The server 110 may be an independent physical server, or a server cluster or distributed system consisting of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, security services, and CDN.

[0030] The terminal 120 may be implemented in hardware or software. For example, when the terminal 120 is implemented in hardware, it may be various electronic devices having a display screen and supporting page display, including but not limited to smart phones, tablet computers, e-book readers, laptop portable computers, and desktop computers, etc. When the terminal 120 device is implemented in software, it may be installed in the electronic devices listed above; it may be implemented as multiple software or software modules (such as software or software modules used to provide distributed services), or it may be implemented as a single software or software module, which is not specifically limited here.

[0031] It should be noted that the address recommendation method provided in the embodiment of the present application can be executed by the terminal 120 or by the server 110. It should be understood that Figure 1 The number of terminals, networks and servers in the embodiment is only for illustration and is not intended to limit the number of terminals, networks and servers. Any number of terminals, networks and servers may be provided as required.

[0032] Figure 2 FIG. 2 shows a schematic diagram of the hardware structure of an exemplary electronic device 200 provided in an embodiment of the present disclosure. Figure 2 As shown, the electronic device 200 may include: a processor 202, a memory 204, a network module 206, a peripheral interface 208 and a bus 210. The processor 202, the memory 204, the network module 206 and the peripheral interface 208 are connected to each other in communication within the electronic device 200 through the bus 210.

[0033] Processor 202 may be a central processing unit (CPU), a neural network processor (NPU), a microcontroller (MCU), a programmable logic device, a digital signal processor (DSP), an application specific integrated circuit (ASIC), or one or more integrated circuits. Processor 202 may be used to perform functions related to the technology described in this disclosure. In some embodiments, processor 202 may also include multiple processors integrated into a single logical component. For example, Figure 2 As shown, processor 202 may include a plurality of processors 202a, 202b, and 202c.

[0034] The memory 204 may be configured to store data (eg, instructions, computer code, etc.). Figure 2As shown, the data stored in the memory 204 may include program instructions (e.g., program instructions for implementing the address recommendation method of the embodiment of the present disclosure) and data to be processed (e.g., the memory may store configuration files of other modules, etc.). The processor 202 may also access the program instructions and data stored in the memory 204, and execute the program instructions to operate on the data to be processed. The memory 204 may include a volatile storage device or a non-volatile storage device. In some embodiments, the memory 204 may include a random access memory (RAM), a read-only memory (ROM), an optical disk, a magnetic disk, a hard disk, a solid-state drive (SSD), a flash memory, a memory stick, etc.

[0035] The network module 206 can be configured to provide communication with other external devices to the electronic device 200 via a network. The network can be any wired or wireless network capable of transmitting and receiving data. For example, the network can be a wired network, a local wireless network (e.g., Bluetooth, WiFi, near field communication (NFC) etc.), a cellular network, the Internet or a combination thereof. It is understood that the type of network is not limited to the above specific examples. In some embodiments, the network module 206 can include any number of network interface controllers (NICs), radio frequency modules, transceivers, modems, routers, gateways, adapters, cellular network chips, etc., in any combination.

[0036] The peripheral interface 208 can be configured to connect the electronic device 200 to one or more peripheral devices to achieve information input and output. For example, the peripheral devices can include input devices such as a keyboard, a mouse, a touch pad, a touch screen, a microphone, and various sensors, and output devices such as a display, a speaker, a vibrator, and an indicator light.

[0037] The bus 210 can be configured to transmit information between various components of the electronic device 200 (e.g., the processor 202, the memory 204, the network module 206, and the peripheral interface 208), such as an internal bus (e.g., a processor-memory bus), an external bus (USB port, PCI-E bus), etc.

[0038] It should be noted that, although the architecture of the electronic device 200 only shows the processor 202, the memory 204, the network module 206, the peripheral interface 208 and the bus 210, in the specific implementation process, the architecture of the electronic device 200 may also include other components necessary for normal execution. In addition, it can be understood by those skilled in the art that the architecture of the electronic device 200 may also only include the components necessary for implementing the embodiments of the present disclosure, and does not necessarily include all the components shown in the figure.

[0039] In the related art, the address recommendation system mainly refers to the location of the located object and the distance of the address during the recall and sorting stages. Since there are fewer reference features in the recall and sorting stages, the accuracy of address recommendation is not high, resulting in a low adoption rate of recommended addresses and the inability to achieve accurate recommendations. Therefore, how to improve the accuracy of address recommendations and the adoption rate of recommended addresses has become a technical problem that needs to be solved urgently.

[0040] In view of this, the embodiments of the present disclosure provide an address recommendation method and related devices. By obtaining the target location of the positioning object and combining the address popularity information in the address database, preliminary screening is performed to obtain candidate addresses. The candidate addresses are then sorted to determine the target address, and the target address is output, which can improve the accuracy and adoption rate of address recommendations.

[0041] See also Figure 3 , Figure 3 The schematic flow chart of the address recommendation method according to the embodiment of the present disclosure is shown. The address recommendation method according to the embodiment of the present disclosure can be deployed on a server or a terminal. Figure 3 In the method 300 for address recommendation, the method 300 for address recommendation may further include the following steps.

[0042] In step S310, the target position is acquired.

[0043] The target location may refer to a location related to the positioning object, for example, it may be real-time positioning information or an input address query word.

[0044] In some embodiments, obtaining the target location includes:

[0045] Acquire the positioning information of the positioning object to obtain the target position;

[0046] or,

[0047] The input address text is detected, address information of the lowest address level in the address text is extracted, and the target location is obtained based on the address information of the lowest address level.

[0048] Among them, obtaining the target position of the positioning object can obtain the real-time positioning information of the positioning object through electronic devices (such as mobile phones, tablets, computers, etc.). It is also possible to detect the input address text and extract the address information of the lowest address level from the address text as the target location. The addresses can be graded according to the degree of detail of the address. The lower the address level, the more detailed the address information. For example, the address "AA District, BB Street, CC, DD Building, EE Floor, Room FF" has the lowest address level and the highest address level is AA District. In the related art, location information is extracted by character matching, which is prone to misrecommendations. In contrast, in the embodiment of the present disclosure, the later levels are extracted by address grading, which can effectively identify more effective target addresses. When the input address text is detected, the lower levels can be used as features in the input model.

[0049] In step S320, a candidate address is determined from the addresses based on the target location and address heat information of the addresses in the address database; wherein the address heat information is used to indicate the popularity of the address.

[0050] Among them, the address popularity information is used to indicate the popularity of the address, which can be determined based on at least one of the number of clicks (for example, the number of clicks within a period of time), the number of recommendations, and the number of searches. The address popularity information can be determined based on weighted average, exponential decay, comprehensive scoring, etc., which are not limited here. Determining the candidate address based on the target location and the address popularity information in the address database not only takes into account the distance factor, but also combines the popularity of the address to ensure that the recommended address is more in line with the needs, thereby improving the adoption rate of the recommended address.

[0051] In some embodiments, determining a candidate address from among the addresses based on the target location and address heat information of addresses in an address database includes:

[0052] Calculating the distance between the target location and the address;

[0053] Extracting features of the distance to obtain a distance feature, and obtaining a heat feature based on the heat information of the address;

[0054] Scoring the address based on the distance feature and the heat feature to obtain a recall score;

[0055] The address whose recall score meets a preset condition is determined as the candidate address.

[0056] The preset conditions may include conditions for screening candidate addresses, such as being greater than or equal to a preset value, or recalling the first several with the highest scores. Figure 4 , Figure 4 A schematic diagram of an address recommendation method according to an embodiment of the present disclosure is shown. Figure 4 In the front-end page of the application, interactive operations can be performed. For example, by triggering a control for recommending addresses, such as an automatic positioning button, in the front-end page, an address recommendation operation can be performed. The target location information adress0 can be obtained, and the target location information adress0 can be input into the trained recall model. The recall model extracts features of the target location information adress0 to obtain the target location feature feature_adress0; and calculates the distance dis between the target location feature feature_adress0 and the address feature feature_POI in the feature library POI-ES corresponding to the address database. The recall model scores the addresses in the address database based on the distance dis and the historical usage information of the address feature feature_POI to obtain a recall score. The recall scores are sorted from high to low, and the addresses corresponding to the first n address features feature_POI with the highest recall scores are determined as candidate addresses, where n is a positive integer.

[0057] In some embodiments, scoring the address based on the distance feature and the heat feature to obtain a recall score includes:

[0058] Using the trained recall model to score based on the distance feature and the corresponding distance weight, as well as the heat feature and the corresponding heat weight, to obtain the recall score;

[0059] The trained recall model is obtained by training based on recall training samples, and the recall training samples are obtained based on historical data of clicks on the recommended target address.

[0060] Among them, the historical data of clicks on the recommended target address may refer to key information such as the time when the target address was clicked, the IP address, the number of clicks, etc. The service log data may be a detailed document that records events, errors, warnings, and operation information that occur during the operation of the system or service. Specifically, the time alignment of the historical data and the service log data may be performed to ensure that the timestamps in the historical data and the service log are synchronized so that the click events can be accurately matched with the log records. It is possible to analyze whether the request type (such as GET, POST, etc.) corresponding to the click event in the historical data is consistent with the request type recorded in the service log to determine whether the click event is actually received and processed by the system. The aligned data can ensure the validity of the historical data, thereby improving the accuracy of the recall model.

[0061] Furthermore, geographic calculation methods (such as the Haversine formula) can be used to calculate the distance between the target location and each address in the address database. The calculated distance is subjected to feature extraction, such as normalization processing, segmentation processing, etc., to meet the requirements of the model input. For example, the distance between each address and the target location can be calculated based on the longitude and latitude of the target location and the longitude and latitude of each address in the address database. And the distance is normalized to the [0,1] interval, and divided into multiple intervals, such as 5-10km, etc. The address heat information of each address is obtained from the address database and feature extraction is performed, such as normalization processing, segmentation processing, etc. The trained recall model is used to score the distance feature and the corresponding distance weight, as well as the heat feature and the corresponding heat weight to obtain the recall score. The recall model can be a logistic regression model, a random forest model, or other machine learning model. The top N addresses with the highest recall scores are determined as candidate addresses.

[0062] In step S330, the candidate addresses are sorted and scored based on a preset dimension to obtain a sorting score.

[0063] Among them, the sorting and scoring can be based on one or more preset dimensions. Feature fusion through more dimensional features can improve the recommendation effect. Specifically, fusion can be based on regional features and fence features (block information), such as whether the target location and address are in the same fence. In a specific area, due to the dense buildings, address recommendations should be more sensitive to distance, and the use of regional features can better alleviate this problem. Sometimes the target location and address are very close, but located in different fences. At this time, addresses within the same fence should be recommended first. For address priority features, addresses with higher priority can be considered when recommending addresses. For example, addresses near the permanent residence can be recommended. Sorting can also be performed based on address category, level, data quality and other features to improve the recommendation effect of the model.

[0064] In some embodiments, the candidate addresses are sorted and scored based on a preset dimension, including:

[0065] Scoring the candidate addresses for each of the preset dimensions to obtain an intermediate score;

[0066] Obtaining the ranking score based on the intermediate score under each preset dimension and the corresponding preset dimension weight;

[0067] The scoring of the candidate addresses for each of the preset dimensions to obtain an intermediate score includes at least one of the following:

[0068] The preset dimension includes a region dimension, and the ranking score when the candidate address and the target location are located in the same region is greater than the ranking score when the candidate address and the target location are located in different regions;

[0069] The preset dimension includes an address priority dimension, the ranking score when the candidate address is within a first preset range of the first address is greater than the ranking score when the candidate address is within a second preset range of the second address; the address priority of the first address is higher than that of the second address;

[0070] The preset dimension includes an address type dimension, and the ranking score when the candidate address and the target location belong to the same address type is greater than the ranking score when the candidate address and the target location belong to different address types;

[0071] The preset dimension includes an address level dimension, and the lower the address level of the candidate address, the higher the corresponding ranking score;

[0072] The preset dimension includes an address data quality dimension, and the higher the address data quality of the candidate address is, the higher the corresponding ranking score is.

[0073] The preset dimensions include at least one of the regional dimension, the address priority dimension, the address type dimension, the address level dimension, and the address data quality dimension. Each dimension can have a corresponding scoring rule to determine the correlation between the candidate address and the target location. One or more preset dimensions can more comprehensively evaluate the correlation between the candidate address and the target location, thereby improving the accuracy of address matching and helping to find the address that best meets the needs more quickly. It can provide more accurate address matching and sorting, and enhance the flexibility and convenience of address recommendation.

[0074] For example, if the target location is a commercial area S in YY District, XX City, where the positioning object is currently located, the candidate addresses can be sorted and scored according to the preset dimensions. Address A: AA, a residential area near ZZ Road Pedestrian Street, YY District, XX City; Address B: Building 2 in BB, a residential area near ZZ' Street, YY' District, XX City. Based on the scoring of the regional dimension, if address A and the target location are in the same area, the median score of address A under the regional dimension is 10 points. If address B and the target location are in different areas, the median score of address B under the regional dimension is 5 points. Based on the scoring of the address priority dimension, the address priority of ZZ Road Pedestrian Street is higher than that of ZZ' Street. Address A is near ZZ Road Pedestrian Street (a high-priority commercial area) (i.e., within the preset range), and address B is near ZZ' Street. The median score of address A under the address priority dimension is 8 points, and the median score of address B is 6 points. Address type dimension: If both address A and address B are in residential areas, the median scores of address A and address B under the address type dimension are both 8 points. Address level dimension: Address A is specific to residential area AA, and address B is specific to building number. The median score of address A under the address level dimension is 5 points, and the median score of address B is 8 points. Address data quality dimension: The data of address A and address B both include region, type, level, etc., and the data information is complete and accurate. The median scores of address A and address B under the address data quality dimension are both 9 points. The weight corresponding to the regional dimension can be α1, the weight corresponding to the address priority dimension can be α2, the weight corresponding to the address type dimension can be α3, the weight corresponding to the address level dimension can be α4, and the weight corresponding to the address data quality dimension can be α5. Then the ranking score of address A = 10×α1+8×α2+8×α3+5×α4+9×α5; the ranking score of address B = 5×α1+6×α2+8×α3+8×α4+9×α5. It should be understood that this example is only for explanation and is not intended to limit the number of preset dimensions and scoring rules. The preset dimensions and scoring rules can be set as needed and are not limited here.

[0075] In some embodiments, the candidate addresses are sorted and scored based on a preset dimension to obtain a sorting score, including:

[0076] Extracting features corresponding to a plurality of the preset dimensions from the candidate address to obtain a plurality of candidate address features under the plurality of the preset dimensions;

[0077] Using the trained ranking model to score the candidate address features under each preset dimension to obtain an intermediate score;

[0078] The ranking score of the candidate address is obtained by fusing the intermediate score under each preset dimension with the corresponding preset dimension weight.

[0079] Among them, by extracting the features of the candidate address under multiple preset dimensions, the trained sorting model is used to score the features under each preset dimension to obtain the intermediate score, and then the intermediate score under each preset dimension is fused with the corresponding preset dimension weight to finally obtain the ranking score of the candidate address. This method not only takes into account the features of multiple dimensions, but also combines the predictive ability of the model to ensure that the recommended address is more accurate, thereby improving the quality and adoption rate of the recommendation. For example, for candidate address A, it extracts features under the preset dimensions to obtain candidate features, and uses the trained sorting model to score these features to obtain the intermediate scores: distance feature: 0.5, address priority feature: 0.8, fence feature: 0.6. The intermediate score under each preset dimension and its corresponding preset dimension weight, multiply the intermediate score under each preset dimension by its corresponding preset dimension weight. Add the weighted intermediate scores to obtain the final ranking score. Assume that the preset dimension weights are as follows: distance weight: 0.3, address priority weight: 0.4. Fence weight: 0.2, then the final ranking score is calculated as (0.5×0.3)+(0.8×0.4)+(0.6×0.2)=0.59.

[0080] In some embodiments, the preset dimension includes a region dimension, and the intermediate score when the candidate address and the target location are located in the same region is greater than the intermediate score when the candidate address and the target location are located in different regions.

[0081] In some embodiments, the preset dimension may include a fence feature dimension. The fence feature may be extracted from the target location information adress0 to obtain the target location fence feature block_feature_adress0. It may be determined whether the target location fence feature block_feature_adress0 and the address feature feature_POI are in the same fence area, and the candidate addresses in the same fence area may be preferentially recommended.

[0082] In some embodiments, the preset dimension includes an address priority dimension, the median score of the candidate address when it is within a first preset range of the first address is greater than the median score of the candidate address when it is within a second preset range of the second address; the address priority of the first address is higher than that of the second address.

[0083] Among them, by setting the priorities of different addresses, and calculating the intermediate scores of candidate addresses based on the relative positions between the candidate addresses and these addresses. By comparing the intermediate scores in different address ranges, it is possible to determine which address has a greater impact on the candidate address, and to effectively select a more preferred option from multiple candidate addresses, ensuring that the high-priority address has a greater weight in the evaluation, so that the final result is more in line with expectations. For example, a resident address can be an address that meets a preset priority, or the address priority of a resident address is higher than the address priority of a non-resident address.

[0084] In some embodiments, the preset dimension includes an address type dimension, and the intermediate score when the candidate address and the target location belong to the same address type is greater than the intermediate score when the candidate address and the target location belong to different address types.

[0085] In some embodiments, the preset dimension includes an address level dimension, and the lower the address level of the candidate address, the higher the corresponding intermediate score. Among them, the lower the address level, the more detailed the address is, and the more accurate the location information can be provided; on the contrary, if the address level is higher, it means that the address is more general and may not be able to accurately locate the specific location. For example, if address A1 is "AA City BB District", and address A2 is "AA City BB District CC Street DD Road EE", address A2 provides more detailed location information, and its corresponding intermediate score is higher than the intermediate score of address A1. In some embodiments, the preset dimension includes an address data quality dimension, and the higher the address data quality of the candidate address, the higher the corresponding intermediate score. The address data quality dimension may refer to the completeness and accuracy of address information.

[0086] In some embodiments, method 300 further includes:

[0087] The initial model is trained based on the sorting training samples to obtain the trained sorting model, including:

[0088] The sorted training samples are obtained based on the historical data of clicks on the recommended target addresses, the selected addresses among the target addresses are used as positive sample data, and the unselected addresses among the target addresses are used as negative sample data;

[0089] Performing feature extraction on the positive sample data and the negative sample data to obtain sorted positive sample features and sorted negative sample features under the preset dimension;

[0090] Obtaining a loss function based on the sorted positive sample features and the sorted negative sample features;

[0091] The model parameters of the initial model are adjusted to minimize the loss function, thereby obtaining the trained sorting model.

[0092] The history data of clicks on the recommended target address may include the first address selected and / or the second address saved in the target address. The service log data may record various operations on the target address in the system, such as searching, clicking, saving, etc. Figure 4 As shown, a sorting training sample can be obtained based on historical data and service log data, including a selected or saved first address and an unselected second address. The selected and / or saved first address is used as positive sample data, and the unselected second address in the target address is used as negative sample data. The preset dimension may include, for example, at least one of distance, heat, building density, road network, functional area division, POI density, fence information, address category, address level, data quality, etc. After feature extraction based on the preset dimension, sorting positive sample features and sorting negative sample features can be obtained. A loss function, such as a binary cross-entropy loss function (Binary Cross-EntropyLoss), and an optimization algorithm such as a gradient descent method and an Adam optimizer can be used to adjust the model parameters to minimize the loss function, and the above process is repeated until the model converges. In this way, the initial model can be effectively trained based on historical data and service log data to obtain a trained sorting model. Not only the historical data of the address in the application is considered, but also multiple source features are combined for fusion sorting, which improves the accuracy and adoption rate of address recommendations.

[0093] See also Figure 5 , Figure 5 A schematic diagram of a sorting model according to an embodiment of the present disclosure is shown. Figure 5 In the example, the sorting model can be a wide&deep model. The candidate address features can be input into the deep side and the shallow side to obtain the deep side features and the shallow side features. The shallow side features directly act on the output layer of the sorting model, which can provide a stronger memory ability and improve the accuracy of address recommendation.

[0094] In step S340, a preset number of the candidate addresses with the largest ranking scores are determined as target addresses corresponding to the target positions, and the target addresses are output.

[0095] In this case, the recommended target addresses can be displayed, for example in the form of a list.

[0096] It can be seen that according to the address recommendation method of the embodiment of the present disclosure, the target location of the positioning object is obtained and combined with the address popularity information in the address database to perform preliminary screening to obtain candidate addresses. The candidate addresses are then sorted to determine the target address, and the target address is output. This can improve the accuracy and adoption rate of address recommendations.

[0097] It should be noted that the method of the embodiment of the present disclosure can be performed by a single device, such as a computer or a server. The method of the present embodiment can also be applied in a distributed scenario and completed by multiple devices cooperating with each other. In the case of such a distributed scenario, one of the multiple devices can only perform one or more steps in the method of the embodiment of the present disclosure, and the multiple devices will perform address recommendations with each other to complete the method described.

[0098] It should be noted that the above describes some embodiments of the present disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the above embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0099] Based on the same technical concept, corresponding to any of the above-mentioned embodiments and methods, the present disclosure also provides an address recommendation device, see Figure 6 , the address recommendation device, the device comprising:

[0100] An acquisition module is used to obtain a target position;

[0101] A recall module, configured to determine a candidate address from the addresses based on the target location and address heat information of the addresses in the address database; wherein the address heat information is used to indicate the attention level of the address;

[0102] A sorting module, used to sort and score the candidate addresses based on a preset dimension to obtain a sorting score;

[0103] The recommendation module is used to determine a preset number of the candidate addresses with the largest ranking scores as target addresses corresponding to the target position, and output the target addresses.

[0104] For the convenience of description, the above device is described by dividing it into various modules according to its functions. Of course, when implementing the present disclosure, the functions of each module can be implemented in the same or multiple software and / or hardware.

[0105] The device of the above embodiment is used to implement the corresponding address recommendation method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be described in detail here.

[0106] Based on the same technical concept, corresponding to any of the above-mentioned embodiment methods, the present disclosure also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the address recommendation method described in any of the above embodiments.

[0107] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0108] The computer instructions stored in the storage medium of the above embodiments are used to enable the computer to execute the address recommendation method described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0109] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present disclosure (including the claims) is limited to these examples. Based on the concept of the present disclosure, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present disclosure as described above, which are not provided in detail for the sake of simplicity.

[0110] In addition, to simplify the description and discussion, and in order not to make the embodiments of the present disclosure difficult to understand, the known power / ground connections to the integrated circuit (IC) chips and other components may or may not be shown in the provided figures. In addition, the device can be shown in the form of a block diagram to avoid making the embodiments of the present disclosure difficult to understand, and this also takes into account the fact that the details of the implementation of these block diagram devices are highly dependent on the platform on which the embodiments of the present disclosure will be implemented (that is, these details should be fully within the scope of understanding of those skilled in the art). Where specific details (e.g., circuits) are set forth to describe exemplary embodiments of the present disclosure, it is apparent to those skilled in the art that the embodiments of the present disclosure can be implemented without these specific details or with changes in these specific details. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0111] Although the present disclosure has been described in conjunction with specific embodiments of the present disclosure, many replacements, modifications and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may use the embodiments discussed.

[0112] The embodiments of the present disclosure are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present disclosure should be included in the scope of protection of the present disclosure.

Claims

1. An address recommendation method, comprising: Get the target location; Determining a candidate address from the addresses based on the target location and address heat information of addresses in an address database; wherein the address heat information is used to indicate the attention level of the address; Sorting and scoring the candidate addresses based on a preset dimension to obtain a ranking score; A preset number of the candidate addresses with the largest ranking scores are determined as target addresses corresponding to the target positions, and the target addresses are output.

2. The method according to claim 1, wherein: Determining a candidate address from among the addresses based on the target location and address heat information of the addresses in the address database includes: Calculating the distance between the target location and the address; Extracting features of the distance to obtain a distance feature, and obtaining a heat feature based on the heat information of the address; Scoring the address based on the distance feature and the heat feature to obtain a recall score; The address whose recall score meets a preset condition is determined as the candidate address.

3. The method according to claim 2, wherein: The address is scored based on the distance feature and the heat feature to obtain a recall score, including: Using the trained recall model to score based on the distance feature and the corresponding distance weight, as well as the heat feature and the corresponding heat weight, to obtain the recall score; The trained recall model is obtained by training based on recall training samples, and the recall training samples are obtained based on historical data of clicks on the recommended target address.

4. The method according to claim 1, wherein: Get the target location, including: Acquire the positioning information of the positioning object to obtain the target position; or, The input address text is detected, address information of the lowest address level in the address text is extracted, and the target location is obtained based on the address information of the lowest address level.

5. The method according to claim 1, wherein: The candidate addresses are sorted and scored based on the preset dimensions to obtain a sorting score, including: Extracting features corresponding to a plurality of the preset dimensions from the candidate address to obtain a plurality of candidate address features under the plurality of the preset dimensions; Using the trained ranking model to score the candidate address features under each preset dimension to obtain an intermediate score; The ranking score of the candidate address is obtained by fusing the intermediate score under each preset dimension with the corresponding preset dimension weight.

6. The method according to claim 1, ranking and scoring the candidate addresses based on a preset dimension, comprises: Scoring the candidate addresses for each of the preset dimensions to obtain an intermediate score; Obtaining the ranking score based on the intermediate score under each preset dimension and the corresponding preset dimension weight; The scoring of the candidate addresses for each of the preset dimensions to obtain an intermediate score includes at least one of the following: The preset dimension includes a region dimension, and the intermediate score when the candidate address and the target location are located in the same region is greater than the intermediate score when the candidate address and the target location are located in different regions; The preset dimension includes an address priority dimension, the median score when the candidate address is within a first preset range of the first address is greater than the median score when the candidate address is within a second preset range of the second address; the address priority of the first address is higher than that of the second address; The preset dimension includes an address type dimension, and the intermediate score when the candidate address and the target location belong to the same address type is greater than the intermediate score when the candidate address and the target location belong to different address types; The preset dimension includes an address level dimension, and the lower the address level of the candidate address, the higher the corresponding intermediate score; The preset dimension includes an address data quality dimension, and the higher the address data quality of the candidate address is, the higher the corresponding intermediate score is.

7. The method according to claim 5, further comprising: The initial model is trained based on the sorting training samples to obtain the trained sorting model, including: The sorted training samples are obtained based on the historical data of clicks on the recommended target addresses, the selected addresses among the target addresses are used as positive sample data, and the unselected addresses among the target addresses are used as negative sample data; Performing feature extraction on the positive sample data and the negative sample data to obtain sorted positive sample features and sorted negative sample features under the preset dimension; Obtaining a loss function based on the sorted positive sample features and the sorted negative sample features; The model parameters of the initial model are adjusted to minimize the loss function, thereby obtaining the trained sorting model.

8. An address recommendation device, comprising: An acquisition module is used to obtain a target position; A recall module, configured to determine a candidate address from the addresses based on the target location and address heat information of the addresses in the address database; wherein the address heat information is used to indicate the attention level of the address; A sorting module, used to sort and score the candidate addresses based on a preset dimension to obtain a sorting score; The recommendation module is used to determine a preset number of the candidate addresses with the largest ranking scores as target addresses corresponding to the target position, and output the target addresses.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the address recommendation method according to any one of claims 1 to 7 when executing the program.

10. A non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause the computer to execute the address recommendation method according to any one of claims 1 to 7.

Images