Method and device for determining popularity of point of interest, equipment, medium and program product

Abstract

The embodiment of the invention provides a point-of-interest popularity determination method and device, equipment, a medium and a program product, and particularly relates to the technical field of big data analysis. The method comprises the following steps: for any interest point in a target area, extracting category distribution characteristics and popularity characteristics of interest points around the interest point on different spatial scales, and respectively combining the category distribution characteristics and the popularity characteristics of each spatial scale to obtain neighborhood characteristics of the interest point on different spatial scales; and based on the extracted neighborhood features, determining an initial popularity value of a target point of interest in the target region by using a clustering algorithm or a popularity prediction model obtained by training, the target point of interest being a newly added point of interest or a point of interest with a missing popularity value. The method is used for achieving the effect of improving the recommendation accuracy and user satisfaction of the map search system.

Description

Technical Field

[0001] This application relates to the field of big data analysis technology, and in particular, to a method, device, equipment, medium and program product for determining the popularity of points of interest. Background Art

[0002] The popularity value of a point of interest (POI) is a quantitative indicator that measures the popularity or attention of the POI. It is usually calculated based on a variety of data and factors, reflecting the degree of interest of users or the public in the POI.

[0003] In practical applications, a map search system may perform intelligent recommendation and sorting based on the user's search intent and context information, combined with the popularity value of the POI. For example, when a user searches for "restaurant", the system may sort according to factors such as the popularity value of the restaurant, user reviews, distance, etc. to provide results that better meet the user's needs.

[0004] However, for newly added POIs, since they have just been added to the map search system and relevant user behavior data, geographical location factors, etc. have not been fully accumulated, their popularity values are usually missing. Summary of the Invention

[0005] Embodiments of this application provide a method, device, equipment, medium and program product for determining the popularity of points of interest, so as to provide a reasonable popularity estimate for newly added POIs or POIs with missing popularity values, thereby improving the recommendation accuracy and user satisfaction of the map search system.

[0006] In a first aspect, embodiments of this application provide a method for determining the popularity of a point of interest, including:

[0007] For any point of interest in the target area, at different spatial scales, extract the category distribution characteristics and popularity characteristics of the surrounding points of interest, and combine the category distribution characteristics and popularity characteristics of each spatial scale respectively to obtain the neighborhood characteristics of this point of interest at different spatial scales;

[0008] Based on the extracted neighborhood characteristics, use a clustering algorithm or a trained popularity prediction model to determine the initial popularity value of the target point of interest in the target area, where the target point of interest is a newly added point of interest or a point of interest with a missing popularity value.

[0009] In a possible implementation manner, the step of, for any point of interest in the target area, at different spatial scales, extracting the category distribution characteristics and popularity characteristics of the surrounding points of interest, and combining the category distribution characteristics and popularity characteristics of each spatial scale respectively to obtain the neighborhood characteristics of this point of interest at different spatial scales includes:

[0010] For any point of interest in the target area, identify the surrounding points of interest of the point of interest at each defined spatial scale. For the surrounding points of interest identified at each spatial scale, count their category distribution information to generate category distribution features, and extract heat features based on their heat information, obtaining the category distribution features and heat features corresponding to the point of interest at different spatial scales respectively. Concatenate the category distribution features and heat features of each spatial scale respectively to generate the neighborhood features of the point of interest at different spatial scales. The category distribution information includes the number or proportion of different category points of interest among the surrounding points of interest, and the heat information includes the heat values of multiple points of interest among the surrounding points of interest.

[0011] In a possible implementation manner, based on the extracted neighborhood features, using a clustering algorithm or a trained heat prediction model to determine the initial heat value of the target point of interest in the target area includes:

[0012] Based on the extracted neighborhood features, perform clustering processing on the points of interest in the target area, and determine the initial heat value of the target point of interest in the target area based on the clustering result; or,

[0013] Input the neighborhood features of the target point of interest into the trained heat prediction model for point of interest heat prediction to obtain the initial heat value of the target point of interest. The heat prediction model is trained using the neighborhood features of the sample points of interest in the target area and their corresponding heat values as training samples, and the sample points of interest are the points of interest in the target area whose heat values meet the preset first heat value condition.

[0014] In a possible implementation manner, based on the extracted neighborhood features, perform clustering processing on the points of interest in the target area, and determine the initial heat value of the target point of interest in the target area based on the clustering result, including:

[0015] Based on the extracted neighborhood features, divide the points of interest in the target area into different point of interest clusters, and the points of interest in the same point of interest cluster have similarity in neighborhood features;

[0016] For any target point of interest in the target area, determine the target point of interest cluster to which the target point of interest belongs, identify a reference point of interest from the target point of interest cluster, and determine the initial heat value of the target point of interest based on the heat value of the reference point of interest. The reference point of interest is the point of interest in the target point of interest cluster whose heat value meets the preset second heat value condition.

[0017] In a possible implementation manner, based on the extracted neighborhood features, dividing the points of interest in the target area into different point of interest clusters includes:

[0018] Based on the extracted neighborhood features and other attribute features of the interest points in the target area, calculate the similarity between the interest points in the target area, where the other attribute features are a set of features of the interest points in multiple attribute dimensions;

[0019] Based on the similarity between the interest points in the target area, divide the interest points in the target area into different interest point clusters.

[0020] In a possible implementation manner, the inputting the neighborhood features of the target interest point into the trained popularity prediction model for interest point popularity prediction to obtain the initial popularity value of the target interest point includes:

[0021] For any spatial scale, fuse the neighborhood features of this spatial scale with the other attribute features of the target interest point to generate a fused feature for this spatial scale, and input the fused feature for this spatial scale into the popularity prediction model corresponding to this spatial scale for interest point popularity prediction to obtain the popularity prediction result corresponding to this spatial scale;

[0022] Based on the popularity prediction results of different spatial scales, determine the initial popularity value of the target interest point.

[0023] In a possible implementation manner, the inputting the neighborhood features of the target interest point into the trained popularity prediction model for interest point popularity prediction to obtain the initial popularity value of the target interest point includes:

[0024] Fuse the neighborhood features of different spatial scales of the target interest point with the other attribute features of the target interest point to generate an overall fused feature;

[0025] Input the overall fused feature into the trained popularity prediction model for interest point popularity prediction, and determine the popularity prediction value in the output popularity prediction result as the initial popularity value of the target interest point.

[0026] In a possible implementation manner, the method further includes:

[0027] For any interest point in the target area, convert its multiple types of attribute information into corresponding numerical features, and generate other attribute features of this interest point according to the numerical features corresponding to different attribute information, where the other attribute features are attribute features other than neighborhood features;

[0028] Wherein, the converting its multiple types of attribute information into corresponding numerical features and generating other attribute features of this interest point according to the numerical features corresponding to different attribute information includes:

[0029] Input its basic attribute information into a pre-trained text embedding model to convert it into a first numerical feature, where the basic attribute information includes one or more of the name, category, and region to which the point of interest belongs;

[0030] Based on a preset encoding method, convert its grade attribute information into a second numerical feature, where the grade attribute information reflects the grade of the point of interest on a specific evaluation dimension;

[0031] Convert its distance attribute information into a third numerical feature, where the distance attribute information includes one or more of the relative distance between the point of interest and a public transportation hub, and the relative distance between the point of interest and a commercial activity center;

[0032] Generate other attribute features of the point of interest based on the first numerical feature, the second numerical feature, and the third numerical feature.

[0033] In a second aspect, an embodiment of the present application provides a device for determining the popularity of a point of interest, including:

[0034] An extraction module, configured to extract the category distribution feature and popularity feature of the surrounding points of interest of any point of interest in the target area at different spatial scales, and combine the category distribution feature and popularity feature of each spatial scale respectively to obtain the neighborhood feature of the point of interest at different spatial scales;

[0035] A determination module, configured to determine the initial popularity value of the target point of interest in the target area based on the extracted neighborhood feature by using a clustering algorithm or a trained popularity prediction model, where the target point of interest is a newly added point of interest or a point of interest with a missing popularity value.

[0036] In a possible implementation manner, the extraction module is specifically configured to:

[0037] For any point of interest in the target area, identify the surrounding points of interest of the point of interest at each defined spatial scale. For the surrounding points of interest identified at each spatial scale, count its category distribution information to generate a category distribution feature, and extract a popularity feature based on its popularity information, to obtain the category distribution feature and popularity feature corresponding to the point of interest at different spatial scales respectively. Concatenate the category distribution feature and popularity feature of each spatial scale respectively to generate the neighborhood feature of the point of interest at different spatial scales, where the category distribution information includes the number or proportion of different category points of interest among the surrounding points of interest, and the popularity information includes the popularity values of multiple points of interest among the surrounding points of interest.

[0038] In a possible implementation manner, the determination module is specifically configured to:

[0039] Based on the extracted neighborhood features, perform clustering processing on the points of interest in the target area, and determine the initial heat value of the target points of interest in the target area based on the clustering result; or,

[0040] Input the neighborhood features of the target point of interest into the trained heat prediction model for predicting the heat of the point of interest, and obtain the initial heat value of the target point of interest. The heat prediction model is trained with the neighborhood features of the sample points of interest in the target area and their corresponding heat values as training samples, and the sample points of interest are the points of interest in the target area whose heat values meet the preset first heat value condition.

[0041] In a possible implementation manner, the determining module is specifically configured to:

[0042] Based on the extracted neighborhood features, divide the points of interest in the target area into different point-of-interest clusters, and the points of interest in the same point-of-interest cluster have similarity in neighborhood features;

[0043] For any target point of interest in the target area, determine the target point-of-interest cluster to which the target point of interest belongs, identify the reference point of interest from the target point-of-interest cluster, and determine the initial heat value of the target point of interest based on the heat value of the reference point of interest. The reference point of interest is the point of interest in the target point-of-interest cluster whose heat value meets the preset second heat value condition.

[0044] In a possible implementation manner, the determining module is specifically configured to:

[0045] Based on the extracted neighborhood features and other attribute features of the points of interest in the target area, calculate the similarity between the points of interest in the target area. The other attribute features are the feature sets of the points of interest in multiple attribute dimensions;

[0046] Based on the similarity between the points of interest in the target area, divide the points of interest in the target area into different point-of-interest clusters.

[0047] In a possible implementation manner, the determining module is specifically configured to:

[0048] For any spatial scale, perform feature fusion on the neighborhood features at this spatial scale and other attribute features of the target point of interest to generate the fusion features for this spatial scale, input the fusion features for this spatial scale into the heat prediction model corresponding to this spatial scale for predicting the heat of the point of interest, and obtain the heat prediction result corresponding to this spatial scale;

[0049] Based on the heat prediction results of different spatial scales, determine the initial heat value of the target point of interest.

[0050] In a possible implementation manner, the determining module is specifically configured to:

[0051] Fuse the neighborhood features of different spatial scales of the target point of interest with other attribute features of the target point of interest to generate an overall fused feature;

[0052] Input the overall fused feature into the trained heat prediction model for predicting the heat of the point of interest, and determine the heat prediction value in the output heat prediction result as the initial heat value of the target point of interest.

[0053] In a possible implementation manner, the heat determination device of the point of interest is further specifically configured to:

[0054] For any point of interest in the target area, convert its various types of attribute information into corresponding numerical features, and generate other attribute features of the point of interest according to the numerical features corresponding to different attribute information, where the other attribute features are attribute features other than neighborhood features;

[0055] Wherein, the heat determination device of the point of interest is further specifically configured to:

[0056] Input its basic attribute information into a pre-trained text embedding model to be converted into a first numerical feature, where the basic attribute information includes one or more of the name, category, and region to which the point of interest belongs;

[0057] Based on a preset coding method, convert its rank attribute information into a second numerical feature, where the rank attribute information reflects the rank of the point of interest in a specific evaluation dimension;

[0058] Convert its distance attribute information into a third numerical feature, where the distance attribute information includes one or more of the relative distance between the point of interest and a public transportation hub and the relative distance between the point of interest and a commercial activity center;

[0059] Generate other attribute features of the point of interest based on the first numerical feature, the second numerical feature, and the third numerical feature.

[0060] In a third aspect, an embodiment of the present application provides a heat determination device for a point of interest, including: a memory, a processor;

[0061] The memory stores computer execution instructions;

[0062] The processor executes the computer execution instructions stored in the memory, so that the processor executes the above first aspect and / or various possible implementation manners of the first aspect.

[0063] Fourthly, an embodiment of the present application provides a computer-readable storage medium storing computer-executable instructions, which are used to implement the first aspect and / or various possible implementation manners of the first aspect as described above when executed by a processor.

[0064] Fifthly, an embodiment of the present application provides a computer program product including a computer program, which implements the first aspect and / or various possible implementation manners of the first aspect as described above when executed by a processor.

[0065] For any point of interest in the target area, the method, device, equipment, medium and program product for determining the popularity of the point of interest provided by the embodiments of the present application can extract the category distribution features and popularity features of the surrounding points of interest at different spatial scales, and combine the category distribution features and popularity features extracted at each spatial scale respectively to obtain the neighborhood features of the point of interest at different spatial scales. By extracting neighborhood features at multiple spatial scales, the environmental features of the point of interest within different spatial scale ranges can be captured. Based on the extracted neighborhood features, a clustering algorithm or a trained popularity prediction model can be used to determine the initial popularity value of the target point of interest in the target area. This multi-scale analysis can avoid the bias that may be brought by a single scale, thereby improving the accuracy of popularity prediction. Through the clustering algorithm or the popularity prediction model, the popularity of newly added points of interest or points of interest with missing popularity values can be estimated quickly. By combining multi-scale feature extraction, clustering analysis and deep learning models, the method can provide a reasonable popularity estimate for newly added points of interest or points of interest with missing popularity values, thereby improving the recommendation accuracy and user satisfaction of the map search system. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] The drawings herein are incorporated into the specification and constitute a part of the specification, showing embodiments consistent with the present application and used to explain the principles of the present application together with the specification.

[0067] Figure 1 It is a schematic diagram of the scenario of the method for determining the popularity of the point of interest provided by the present application;

[0068] Figure 2 It is a schematic flowchart of the method for determining the popularity of the point of interest provided by the present application Figure 1 ;

[0069] Figure 3 It is a schematic flowchart of the method for determining the popularity of the point of interest provided by the present application Figure 2 ;

[0070] Figure 4 It is a schematic diagram of generating the feature expression of the point of interest provided by the present application;

[0071] Figure 5Schematic diagram of the process for estimating popularity through clustering provided by this application;

[0072] Figure 6 Schematic diagram of the process for estimating popularity using multiple models provided by this application;

[0073] Figure 7 Schematic diagram of the process for estimating popularity using a comprehensive model provided by this application;

[0074] Figure 8 Schematic diagram of the structure of the device for determining the popularity of points of interest provided by this application;

[0075] Figure 9 Schematic diagram of the structure of the device for determining the popularity of points of interest provided by this application.

[0076] Through the above-mentioned drawings, specific embodiments of this application have been shown, and there will be more detailed descriptions hereinafter. These drawings and textual descriptions are not intended to limit the scope of the concept of this application in any way, but to illustrate the concept of this application to those skilled in the art by referring to specific embodiments. Detailed implementation manners

[0077] Here, exemplary embodiments will be described in detail, and examples are shown in the drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementation manners described in the following exemplary embodiments do not represent all implementation manners consistent with this application. On the contrary, they are merely examples of devices and methods consistent with some aspects of this application as detailed in the appended claims.

[0078] In a map search system, a major challenge faced by newly added points of interest is the lack of popularity values. This is mainly because these points of interest have just been added to the system, and relevant user behavior data, geographical location factors, etc. have not been fully accumulated and analyzed. Since the newly added points of interest have just been launched, user behavior data such as access and evaluation of them is still in the accumulation stage and is not sufficient to form a stable popularity value. The popularity value is one of the important factors determining the ranking of points of interest in search results. The missing popularity value may cause the newly added points of interest to have a lower ranking in search results and be difficult to be discovered by users.

[0079] The method for determining the popularity of points of interest provided by this application, for any point of interest in the target area, at different spatial scales, can extract the category distribution characteristics and popularity characteristics of the surrounding points of interest, and combine the category distribution characteristics and popularity characteristics extracted at each spatial scale respectively to obtain the neighborhood characteristics of this point of interest at different spatial scales. By extracting neighborhood characteristics at multiple spatial scales, the environmental characteristics of points of interest within different spatial scale ranges can be captured. Based on the extracted neighborhood characteristics, a clustering algorithm or a trained popularity prediction model can be used to determine the initial popularity value of the target point of interest in the target area. This multi-scale analysis can avoid the biases that may be brought by a single scale, thereby improving the accuracy of popularity prediction. Through a clustering algorithm or a popularity prediction model, the popularity of newly added points of interest or points of interest lacking popularity values can be quickly estimated. This method solves the technical problem of lacking popularity values for points of interest by combining multi-scale feature extraction, clustering analysis, and deep learning models.

[0080] Figure 1 is a schematic diagram of the scenario of the method for determining the popularity of points of interest provided by this application, as Figure 1 shown. In this application scenario, the analysis of points of interest data can be performed on the terminal device 101, and the neighborhood characteristics at multiple spatial scales can be extracted for all points of interest in the points of interest data to obtain the neighborhood characteristics corresponding to different spatial scales. The extracted neighborhood characteristics can be processed through a clustering algorithm or a deep learning model. These methods can be used to identify patterns and trends in the data. Based on the extracted neighborhood characteristics, the initial popularity value of newly added or points of interest lacking popularity values is determined.

[0081] The following uses specific embodiments to elaborate in detail on the technical solutions of this application and how the technical solutions of this application solve the above technical problems. These several specific embodiments below can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of this application will be described below in conjunction with the accompanying drawings.

[0082] Figure 2 is a flowchart of the method for determining the popularity of points of interest provided by this application Figure 1 as Figure 2 shown. This method includes:

[0083] S201. For any point of interest in the target area, at different spatial scales, extract the category distribution characteristics and popularity characteristics of the surrounding points of interest, and combine the category distribution characteristics and popularity characteristics at each spatial scale respectively to obtain the neighborhood characteristics of this point of interest at different spatial scales.

[0084] In the embodiments of this application, the neighborhood characteristics at multiple spatial scales can be extracted for the points of interest in the target area to obtain the neighborhood characteristics at different spatial scales.

[0085] There are various ways to divide geographical regions, and these divisions are usually based on different analysis objectives and requirements. The embodiments of this application are to estimate the potential impact and popularity of newly added points of interest or points of interest with missing heat values.

[0086] In the embodiments of this application, the geographical region to be analyzed, that is, the target region, can be determined, the points of interest within the target region can be determined, multiple spatial scales to be analyzed can be determined, and the neighborhood features of the points of interest at different spatial scales can be extracted. The neighborhood features describe the geographical spatial attributes within a certain spatial range around the point of interest and its interaction with the surrounding environment. Specifically, for each point of interest, the category distribution feature and heat feature of the points of interest around it can be extracted. The surrounding points of interest refer to other points of interest existing around a certain point of interest within a specific spatial range. These points of interest can be various types of locations or facilities, having a certain attraction or functionality, and are usually used to describe and analyze the environmental characteristics of the area where a concerned point of interest is located. The extracted category distribution feature reflects the distribution of different category points of interest around the concerned point of interest. The extracted heat feature reflects the access frequency or popularity of multiple points of interest around the concerned point of interest. By separately extracting the category distribution feature and heat feature at different spatial scales and combining them together, the neighborhood features of the point of interest at different spatial scales can be generated.

[0087] S202. Based on the extracted neighborhood features, use a clustering algorithm or a trained heat prediction model to determine the initial heat value of the target point of interest in the target region, where the target point of interest is a newly added point of interest or a point of interest with a missing heat value.

[0088] In one implementation, based on the extracted neighborhood features, a suitable clustering algorithm can be selected to perform clustering processing on the points of interest. The purpose of clustering is to group the points of interest with similar neighborhood features to identify different types of point-of-interest patterns. Based on the clustering result, an initial heat value is assigned to each newly added or heat-value-missing point of interest.

[0089] In another implementation, based on the extracted neighborhood features, a trained heat prediction model can be used for heat estimation to determine the initial heat value of the newly added or heat-value-missing point of interest.

[0090] The method for determining the popularity of a point of interest provided by the embodiments of the present application can, for any point of interest in a target area, extract the category distribution characteristics and popularity characteristics of the surrounding points of interest at different spatial scales. The category distribution characteristics and popularity characteristics extracted at each spatial scale are respectively combined to obtain the neighborhood characteristics of the point of interest at different spatial scales. By extracting neighborhood characteristics at multiple spatial scales, the environmental characteristics of the point of interest within different spatial scale ranges can be captured. Based on the extracted neighborhood characteristics, a clustering algorithm or a trained popularity prediction model can be used to determine the initial popularity value of the target point of interest in the target area. This multi-scale analysis can avoid the bias that may be brought by a single scale, thereby improving the accuracy of popularity prediction. Through a clustering algorithm or a popularity prediction model, the popularity of newly added points of interest or points of interest with missing popularity values can be quickly estimated. By combining multi-scale feature extraction, clustering analysis, and deep learning models, this method can provide a reasonable popularity estimate for newly added points of interest or points of interest with missing popularity values, thereby achieving the effect of improving the recommendation accuracy and user satisfaction of the map search system.

[0091] Figure 3 Schematic flow of the method for determining the popularity of a point of interest provided by the present application Figure 2 , such as Figure 3 shown. Based on the Figure 2 embodiment, the method for determining the popularity of a point of interest will be described in detail. The method includes:

[0092] S301. For any point of interest in the target area, identify the surrounding points of interest of the point of interest at each defined spatial scale. For the surrounding points of interest identified at each spatial scale, count their category distribution information to generate category distribution characteristics, and extract popularity characteristics based on their popularity information, so as to obtain the category distribution characteristics and popularity characteristics corresponding to the point of interest at different spatial scales respectively. Concatenate the category distribution characteristics and popularity characteristics of each spatial scale respectively to generate the neighborhood characteristics of the point of interest at different spatial scales. The category distribution information includes the number or proportion of different category points of interest among the surrounding points of interest, and the popularity information includes the popularity values of multiple points of interest among the surrounding points of interest.

[0093] In the embodiments of the present application, different spatial scales can be determined, such as 500 meters, 1000 meters, 2000 meters, etc., so as to analyze the neighborhood characteristics of the point of interest at these scales. For a certain point of interest, at each spatial scale, identify the surrounding points of interest of the point of interest. For the surrounding points of interest identified within each spatial scale, count the number or proportion of different category points of interest to obtain category distribution information, and obtain the popularity values of different points of interest among the surrounding points of interest to obtain popularity information.

[0094] The category distribution information reflects the quantity and proportion of different categories of points of interest around the point of interest under attention. The point of interest under attention can be any point of interest in the target area. These categories may include restaurants, stores, entertainment facilities, etc. Through the category distribution information, it is possible to understand whether the service facilities around the point of interest under attention are complete, whether the distribution of various facilities is balanced, and whether there is a clustering phenomenon of certain specific types of facilities. Based on the category distribution information, category distribution features can be extracted.

[0095] The popularity information can include the popularity values of multiple points of interest around the point of interest under attention. Additionally, it can also include the total popularity value, which is obtained by accumulating the popularity values of each point of interest around the point of interest under attention. The popularity value is usually calculated based on multiple factors, such as the number of visits, user evaluations, interaction situations, etc. Therefore, the total popularity value can reflect the concentration degree of population activities and the prosperity degree of commercial activities around the point of interest under attention. Based on the popularity information, popularity features can be extracted.

[0096] The category distribution features and popularity features at each spatial scale are respectively concatenated to generate the neighborhood features of the corresponding point of interest at different spatial scales. Among them, for different spatial scales, during the concatenation process, different weights can be assigned to the category distribution features and popularity features respectively, and then the weighted category distribution features and popularity features are concatenated. By repeating the above process of feature extraction and feature fusion at different spatial scales, the neighborhood features of the point of interest at different spatial scales can be obtained.

[0097] In a possible implementation manner, the method for determining the popularity of a point of interest in the embodiments of the present application may specifically further include:

[0098] For any point of interest in the target area, convert its multiple types of attribute information into corresponding numerical features, and generate other attribute features of this point of interest according to the numerical features corresponding to different attribute information. The other attribute features are attribute features other than the neighborhood features;

[0099] Among them, converting its multiple types of attribute information into corresponding numerical features and generating other attribute features of this point of interest according to the numerical features corresponding to different attribute information may specifically include:

[0100] Input its basic attribute information into a pre-trained text embedding model to convert it into a first numerical feature. The basic attribute information includes one or more of the name, category, and region to which the point of interest belongs;

[0101] Based on a preset coding method, convert its rank attribute information into a second numerical feature. The rank attribute information reflects the rank of the point of interest in a specific evaluation dimension;

[0102] Convert its distance attribute information into a third numerical feature, where the distance attribute information includes one or more of the relative distance between the point of interest and the public transportation hub, and the relative distance between the point of interest and the commercial activity center;

[0103] Generate other attribute features of the point of interest based on the first numerical feature, the second numerical feature, and the third numerical feature.

[0104] In this embodiment, in addition to extracting the neighborhood features of the point of interest, other attribute features other than the neighborhood features can also be extracted. By converting different types of attribute information into corresponding numerical features, the feature vector formed by fusing these numerical features is the other attribute feature of the point of interest.

[0105] It is possible to collect various types of attribute information of the point of interest, including basic attribute information, level attribute information, and distance attribute information. Use a pre-trained text embedding model to convert the basic attribute information into a first numerical feature; based on a preset coding method (such as One-Hot coding), convert the level attribute information into a second numerical feature, and the level attribute information can include the level of the region to which the point of interest belongs; convert the distance attribute information into a third numerical feature; fuse the first numerical feature, the second numerical feature, and the third numerical feature to generate other attribute features of the point of interest. Further, the neighborhood features and other attribute features of the point of interest can also be fused to obtain the feature expression of the point of interest. Refer to Figure 4 As shown, it is a schematic diagram of generating the feature expression of the point of interest provided by this application. By characterizing various attribute information of the point of interest, the feature expression of the point of interest is generated. The feature expression is a multi-dimensional vector, and each dimension represents a specific feature of the point of interest.

[0106] S302. Based on the extracted neighborhood features, use a clustering algorithm or a trained popularity prediction model to determine the initial popularity value of the target point of interest in the target area, where the target point of interest is a newly added point of interest or a point of interest with a missing popularity value.

[0107] In a possible implementation manner, based on the extracted neighborhood features, using a clustering algorithm or a trained popularity prediction model to determine the initial popularity value of the target point of interest in the target area may specifically include:

[0108] Based on the extracted neighborhood features, perform clustering processing on the points of interest in the target area, and determine the initial popularity value of the target point of interest in the target area based on the clustering result; or,

[0109] Input the neighborhood features of the target point of interest into the trained popularity prediction model for predicting the popularity of the point of interest, and obtain the initial popularity value of the target point of interest. The popularity prediction model is trained with the neighborhood features of the sample points of interest in the target area and their corresponding popularity values as training samples. The sample points of interest are the points of interest in the target area whose popularity values meet the preset first popularity value condition.

[0110] In one implementation, using the extracted neighborhood features, clustering analysis can be performed on the points of interest in the target area. Clustering analysis is an unsupervised machine learning method used to group data points so that the data points within the same group have high similarity, while the similarity between different groups is low. Through clustering, groups of points of interest with similar features can be identified. For newly added points of interest and points of interest with missing popularity values, initial popularity values can be assigned to these points of interest based on the clustering results.

[0111] In another implementation, the neighborhood features of the target point of interest can be input into the trained popularity prediction model for popularity prediction to obtain the initial popularity value of the target point of interest.

[0112] Among them, the training process of the popularity prediction model includes: selecting points of interest that meet the preset first popularity value condition from the target area, and calling these points of interest sample points of interest. The sample points of interest have known popularity values. Extract the neighborhood features of the sample points of interest, pair the neighborhood features of the sample points of interest with their corresponding popularity values, and construct training samples. Use the training samples for model training so that the popularity prediction model can learn the relationship between the neighborhood features and the popularity values.

[0113] In one possible implementation, based on the extracted neighborhood features, perform clustering processing on the points of interest in the target area, and determine the initial popularity value of the target point of interest in the target area based on the clustering results. Specifically, it can include:

[0114] Based on the extracted neighborhood features, divide the points of interest in the target area into different point-of-interest clusters. The points of interest in the same point-of-interest cluster have similarity in neighborhood features;

[0115] For any target point of interest in the target area, determine the target point-of-interest cluster to which the target point of interest belongs, identify reference points of interest from the target point-of-interest cluster, and determine the initial popularity value of the target point of interest based on the popularity values of the reference points of interest. The reference points of interest are the points of interest in the target point-of-interest cluster whose popularity values meet the preset second popularity value condition.

[0116] Among them, based on the popularity value of reference points of interest, the initial popularity value of the target point of interest is determined. This can be achieved by calculating methods such as the average value, median value, or weighted average of the popularity values of the reference points of interest. Exemplarily, in the target point-of-interest cluster to which the target point of interest belongs, all points of interest with popularity values are found. These points of interest will serve as reference points of interest, and the popularity values of these reference points of interest are collected into a list, for example: [poi_1 popularity, poi_2 popularity, …, poi_n popularity]. If the popularity values of certain reference points of interest are more representative, higher weights can be assigned to these points. Then, the initial popularity value of the target point of interest is calculated by the method of weighted average. Refer to Figure 5 As shown, it is a schematic diagram of the process of estimating popularity through clustering provided by this application. The similarity between points of interest is utilized to infer the popularity value of points of interest lacking historical data.

[0117] In a possible implementation manner, based on the extracted neighborhood features, the points of interest in the target area are divided into different point-of-interest clusters, which may specifically include:

[0118] Based on the extracted neighborhood features and other attribute features of the points of interest in the target area, the similarity between the points of interest in the target area is calculated, and the other attribute features are a set of features of the points of interest in multiple attribute dimensions;

[0119] Based on the similarity between the points of interest in the target area, the points of interest in the target area are divided into different point-of-interest clusters.

[0120] The other attribute features reflect the features of the points of interest in multiple dimensions such as basic attributes, level attributes, and distance attributes.

[0121] In this implementation manner, based on the extracted neighborhood features and other attribute features, the similarity between points of interest can be calculated. Not only can neighborhood features be used, but other attribute features can also be combined to obtain a more comprehensive and accurate similarity measure. This method can better capture the multi-dimensional relationships between points of interest. The similarity measure can use various methods, for example, cosine similarity calculation can be adopted.

[0122] In the embodiments of this application, deep learning methods can also be adopted for estimating the popularity of points of interest.

[0123] In a possible implementation manner, the neighborhood features of the target point of interest are input into the trained popularity prediction model for predicting the popularity of the point of interest, and the initial popularity value of the target point of interest is obtained, which may specifically include:

[0124] For any spatial scale, fuse the neighborhood features of this spatial scale with other attribute features of the target point of interest to generate fused features for this spatial scale, and input the fused features of this spatial scale into the heat prediction model corresponding to this spatial scale to predict the heat of the point of interest, obtaining the heat prediction result corresponding to this spatial scale;

[0125] Based on the heat prediction results of different spatial scales, determine the initial heat value of the target point of interest.

[0126] In this embodiment, different spatial scales correspond to different heat prediction models. For each spatial scale, the neighborhood features of the target point of interest at this spatial scale can be fused with other attribute features of the target point of interest to obtain the fused features of the target point of interest at this spatial scale. The fusion method can adopt the feature splicing method, that is, splicing different feature vectors into a larger vector. Input the fused features of each spatial scale into the corresponding heat prediction model. Each model outputs the heat prediction result corresponding to this spatial scale.

[0127] Based on the heat prediction results of different spatial scales, determine the initial heat value of the target point of interest. This can be achieved by calculating the average value or weighted average, etc.

[0128] Refer to Figure 6 As shown, it is a schematic diagram of the process of using multiple models for heat estimation provided by this application. Since different points of interest may have different sensitivities to spatial scales, by constructing multiple heat prediction models to adapt to different spatial scales, the characteristic influences of different types of points of interest at different scales can be captured more accurately, thereby improving the accuracy and robustness of heat estimation.

[0129] In a possible implementation manner, input the neighborhood features of the target point of interest into the trained heat prediction model to predict the heat of the point of interest, and obtain the initial heat value of the target point of interest. Specifically, it may include:

[0130] Fuse the neighborhood features of different spatial scales of the target point of interest with other attribute features of the target point of interest to generate overall fused features;

[0131] Input the overall fused features into the trained heat prediction model to predict the heat of the point of interest, and determine the heat prediction value in the output heat prediction result as the initial heat value of the target point of interest.

[0132] Refer to Figure 7 As shown, it is a schematic diagram of the process of using a comprehensive model for heat estimation provided by this application. A comprehensive heat prediction model can be trained for heat estimation.

[0133] The method for determining the popularity of points of interest provided by the embodiments of the present application can capture the environmental features of points of interest within different spatial scale ranges by extracting neighborhood features at multiple spatial scales. In one method, clustering processing can be performed based on the extracted neighborhood features, which can effectively group the points of interest in the target area and identify groups of points of interest with similar neighborhood features. The clustering result is used to determine the initial popularity value of newly added or popularity value missing points of interest. In another method, the neighborhood features of the target point of interest can be input into a trained popularity prediction model for predicting the popularity of the point of interest. This model utilizes the relationship between the neighborhood features and popularity values of sample points of interest in historical data, thereby providing a more accurate popularity estimate. This method combines multi-scale feature extraction, clustering analysis, and deep learning models to provide a reasonable popularity estimate for newly added points of interest or points of interest with missing popularity values, thereby improving the recommendation accuracy and user satisfaction of the map search system.

[0134] Figure 8 The following is a schematic structural diagram of the apparatus for determining the popularity of points of interest provided by the present application, as Figure 8 shown, the apparatus 80 for determining the popularity of points of interest provided in this embodiment includes:

[0135] An extraction module 801, configured to, for any point of interest in the target area, extract the category distribution features and popularity features of the surrounding points of interest at different spatial scales, and combine the category distribution features and popularity features of each spatial scale respectively to obtain the neighborhood features of the point of interest at different spatial scales;

[0136] A determination module 802, configured to determine the initial popularity value of the target point of interest in the target area based on the extracted neighborhood features, where the target point of interest is a newly added point of interest or a point of interest with a missing popularity value, using a clustering algorithm or a trained popularity prediction model.

[0137] In a possible implementation manner, the extraction module is specifically configured to:

[0138] For any point of interest in the target area, identify the surrounding points of interest of the point of interest at each defined spatial scale, and for the surrounding points of interest identified at each spatial scale, count its category distribution information to generate category distribution features, and extract popularity features based on its popularity information, to obtain the category distribution features and popularity features respectively corresponding to the point of interest at different spatial scales, and splice the category distribution features and popularity features of each spatial scale respectively to generate the neighborhood features of the point of interest at different spatial scales, where the category distribution information includes the number or proportion of different category points of interest among the surrounding points of interest, and the popularity information includes the popularity values of multiple points of interest among the surrounding points of interest.

[0139] In a possible implementation manner, the determination module is specifically configured to:

[0140] Based on the extracted neighborhood features, perform clustering on the points of interest in the target area, and determine the initial heat value of the target point of interest in the target area based on the clustering result; or,

[0141] Input the neighborhood features of the target point of interest into the trained heat prediction model for predicting the heat of the point of interest, and obtain the initial heat value of the target point of interest. The heat prediction model is trained with the neighborhood features of the sample points of interest in the target area and their corresponding heat values as training samples. The sample points of interest are the points of interest in the target area whose heat values meet the preset first heat value condition.

[0142] In a possible implementation manner, the determining module is specifically configured to:

[0143] Based on the extracted neighborhood features, divide the points of interest in the target area into different point-of-interest clusters, and the points of interest in the same point-of-interest cluster have similarity in neighborhood features;

[0144] For any target point of interest in the target area, determine the target point-of-interest cluster to which the target point of interest belongs, identify the reference point of interest from the target point-of-interest cluster, and determine the initial heat value of the target point of interest based on the heat value of the reference point of interest. The reference point of interest is the point of interest in the target point-of-interest cluster whose heat value meets the preset second heat value condition.

[0145] In a possible implementation manner, the determining module is specifically configured to:

[0146] Based on the extracted neighborhood features and other attribute features of the points of interest in the target area, calculate the similarity between the points of interest in the target area. The other attribute features are the feature sets of the points of interest in multiple attribute dimensions;

[0147] Based on the similarity between the points of interest in the target area, divide the points of interest in the target area into different point-of-interest clusters.

[0148] In a possible implementation manner, the determining module is specifically configured to:

[0149] For any spatial scale, perform feature fusion on the neighborhood features of the spatial scale and other attribute features of the target point of interest to generate the fusion features for the spatial scale, input the fusion features for the spatial scale into the heat prediction model corresponding to the spatial scale for predicting the heat of the point of interest, and obtain the heat prediction result corresponding to the spatial scale;

[0150] Based on the heat prediction results of different spatial scales, determine the initial heat value of the target point of interest.

[0151] In a possible implementation manner, the determining module is specifically configured to:

[0152] Fuse the neighborhood features of the target point of interest at different spatial scales with other attribute features of the target point of interest to generate an overall fused feature;

[0153] Input the overall fused feature into the trained popularity prediction model for predicting the popularity of the point of interest, and determine the popularity prediction value in the output popularity prediction result as the initial popularity value of the target point of interest.

[0154] In a possible implementation, the popularity determination device of the point of interest is specifically further configured to:

[0155] For any point of interest in the target area, convert its multiple types of attribute information into corresponding numerical features, and generate other attribute features of the point of interest according to the numerical features corresponding to different attribute information, where the other attribute features are attribute features other than the neighborhood features;

[0156] Among them, the popularity determination device of the point of interest is specifically further configured to:

[0157] Input its basic attribute information into a pre-trained text embedding model to convert it into a first numerical feature, where the basic attribute information includes one or more of the name, category, and region to which the point of interest belongs;

[0158] Based on a preset encoding method, convert its rank attribute information into a second numerical feature, where the rank attribute information reflects the rank of the point of interest in a specific evaluation dimension;

[0159] Convert its distance attribute information into a third numerical feature, where the distance attribute information includes one or more of the relative distance between the point of interest and a public transportation hub, and the relative distance between the point of interest and a commercial activity center;

[0160] Generate other attribute features of the point of interest based on the first numerical feature, the second numerical feature, and the third numerical feature.

[0161] The popularity determination device of the point of interest provided in this embodiment can execute the method provided in the above method embodiment, and its implementation principle and technical effect are similar, which will not be elaborated here in this embodiment.

[0162] Figure 9 It is a schematic structural diagram of the popularity determination device of the point of interest provided in this application. As Figure 9 shown, the popularity determination device 90 provided in this embodiment includes: at least one processor 901 and a memory 902. Optionally, the device 90 further includes a communication component 903. Among them, the processor 901, the memory 902, and the communication component 903 are connected through a bus.

[0163] In a specific implementation process, at least one processor 901 executes computer-executable instructions stored in a memory 902, so that at least one processor 901 executes the above-mentioned method.

[0164] For the specific implementation process of the processor 901, reference may be made to the above method embodiment, and its implementation principle and technical effects are similar, which will not be elaborated here in this embodiment.

[0165] In the above embodiment, it should be understood that the processor may be a central processing unit (Central Processing Unit, CPU for short), or other general-purpose processors, digital signal processors (Digital Signal Processor, DSP for short), application specific integrated circuits (Application Specific Integrated Circuit, ASIC for short), etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in combination with the invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules in the processor.

[0166] The memory may include a high-speed memory (Random Access Memory, RAM), and may also include a non-volatile memory (Non-volatile Memory, NVM), such as at least one disk memory.

[0167] The bus may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For the sake of convenience in representation, the bus in the drawings of this application is not limited to only one bus or one type of bus.

[0168] This application also provides a computer program product, including a computer program, which implements the above-mentioned method when executed by a processor.

[0169] This application also provides a computer-readable storage medium, in which computer-executable instructions are stored, and when the processor executes the computer-executable instructions, the above-mentioned method is implemented.

[0170] The above-readable storage medium may be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, a magnetic disk, or an optical disk. The readable storage medium may be any available medium accessible by a general-purpose or special-purpose computer.

[0171] An exemplary readable storage medium is coupled to the processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium may also be an integral part of the processor. The processor and the readable storage medium may be located in an application specific integrated circuit (ASIC). Of course, the processor and the readable storage medium may also exist as discrete components in a device.

[0172] The division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other may be indirect couplings or communication connections through some interfaces, devices, or units, and may be in electrical, mechanical, or other forms.

[0173] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0174] In addition, in each embodiment of the present invention, the functional units may be integrated in a processing unit, may exist separately as individual physical units, or two or more units may be integrated in one unit.

[0175] If a function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present invention. The foregoing storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs, etc., all kinds of media that can store program codes.

[0176] Those of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions. The foregoing program can be stored in a computer-readable storage medium. When this program is executed, it executes the steps including the above method embodiments; and the foregoing storage medium includes: ROMs, RAMs, magnetic disks, or optical discs, etc., all kinds of media that can store program codes.

[0177] Finally, it should be noted that: After considering the specification and practicing the invention disclosed herein, those skilled in the art will readily think of other implementation manners of the present invention. The present invention is intended to cover any variations, uses, or adaptations of the present invention, which follow the general principles of the present invention and include common general knowledge or conventional technical means in the technical field not disclosed in the present invention. It is not limited to the exact structures described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present invention is only limited by the appended claims.

Claims

1. A method for determining the popularity of a point of interest, characterized in that: include: For any point of interest in the target area, the category distribution features and heat features of the surrounding points of interest are extracted at different spatial scales, and the category distribution features and heat features of each spatial scale are combined to obtain the neighborhood features of the point of interest at different spatial scales; Based on the extracted neighborhood features, the initial heat value of the target interest point in the target area is determined using a clustering algorithm or a trained heat prediction model, where the target interest point is a newly added interest point or an interest point with a missing heat value.

2. The method according to claim 1, characterized in that For any point of interest in the target area, the category distribution features and heat features of the surrounding points of interest are extracted at different spatial scales, and the category distribution features and heat features of each spatial scale are respectively combined to obtain the neighborhood features of the point of interest at different spatial scales, including: For any point of interest in the target area, identify the surrounding points of interest of the point of interest at each defined spatial scale, and for the surrounding points of interest identified at each spatial scale, count their category distribution information, generate category distribution features, and extract heat features based on their heat information, and obtain the category distribution features and heat features corresponding to the point of interest at different spatial scales; The category distribution features and heat features of each spatial scale are spliced ​​separately to generate the neighborhood features of the interest point at different spatial scales. The category distribution information includes the number or proportion of interest points of different categories in the surrounding interest points, and the heat information includes the heat values ​​of multiple interest points in the surrounding interest points.

3. The method according to any one of claims 1 or 2, characterized in that The step of determining the initial heat value of the target interest point in the target area based on the extracted neighborhood features by using a clustering algorithm or a trained heat prediction model includes: Based on the extracted neighborhood features, clustering is performed on the interest points in the target area, and based on the clustering results, initial heat values ​​of the target interest points in the target area are determined; or, The neighborhood features of the target interest point are input into the trained heat prediction model to perform interest point heat prediction to obtain the initial heat value of the target interest point. The heat prediction model is obtained by training the neighborhood features of sample interest points in the target area and their corresponding heat values ​​as training samples. The sample interest point is an interest point in the target area whose heat value meets a preset first heat value condition.

4. The method according to claim 3, characterized in that The step of clustering the interest points in the target area based on the extracted neighborhood features, and determining the initial heat values ​​of the target interest points in the target area based on the clustering results, includes: Based on the extracted neighborhood features, the points of interest in the target area are divided into different point of interest clusters, and the points of interest in the same point of interest cluster have similarities in neighborhood features; For any target interest point in the target area, determine the target interest point cluster to which the target interest point belongs, identify a reference interest point from the target interest point cluster, and determine the initial heat value of the target interest point based on the heat value of the reference interest point, wherein the reference interest point is an interest point in the target interest point cluster whose heat value satisfies a preset second heat value condition.

5. The method according to claim 4, characterized in that The dividing the points of interest in the target area into different interest point clusters based on the extracted neighborhood features includes: Calculating the similarity between the points of interest in the target area based on the extracted neighborhood features and other attribute features of the points of interest in the target area, wherein the other attribute features are feature sets of the points of interest in multiple attribute dimensions; Based on the similarities between the interest points in the target area, the interest points in the target area are divided into different interest point clusters.

6. The method according to claim 3, characterized in that The step of inputting the neighborhood features of the target interest point into the trained heat prediction model to perform interest point heat prediction to obtain an initial heat value of the target interest point includes: For any spatial scale, the neighborhood features of the spatial scale are fused with other attribute features of the target interest point to generate fused features for the spatial scale, and the fused features of the spatial scale are input into the heat prediction model corresponding to the spatial scale to predict the heat of the interest point, so as to obtain the heat prediction result corresponding to the spatial scale; Based on the heat prediction results of different spatial scales, the initial heat value of the target interest point is determined.

7. The method according to any one of claims 1 or 2, characterized in that The method further comprises: For any point of interest in the target area, its various types of attribute information are converted into corresponding numerical features, and other attribute features of the point of interest are generated according to the numerical features corresponding to different attribute information, wherein the other attribute features are attribute features other than neighborhood features.

8. A device for determining the heat of a point of interest, characterized in that: include: The extraction module is used to extract the category distribution features and heat features of the surrounding interest points at different spatial scales for any interest point in the target area, and to combine the category distribution features and heat features of each spatial scale to obtain the neighborhood features of the interest point at different spatial scales; A determination module is used to determine the initial heat value of the target interest point in the target area based on the extracted neighborhood features and using a clustering algorithm or a trained heat prediction model. The target interest point is a newly added interest point or an interest point with a missing heat value.

9. A device for determining the heat of a point of interest, characterized in that: include: Memory, processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory, so that the processor performs the method according to any one of claims 1 to 7.

10. A computer-readable storage medium / computer program product, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 7 when executed by a processor; and / or, The computer program product comprises a computer program, which implements the method according to any one of claims 1 to 7 when executed by a processor.

Images