A city functional area identification method considering POI homogeneity and structural similarity

Abstract

The application discloses a city functional area identification method considering POI homogeneity and structural similarity, wherein the method comprises the following steps: obtaining POI data, determining a plurality of neighborhoods corresponding to target POI data; sampling the target POI data according to a breadth search algorithm, a depth search algorithm and a preset walk length, so as to generate a walk sequence corresponding to the target POI data in a first-order neighborhood and / or a high-order neighborhood; inputting the walk sequence corresponding to the POI data into a Skip-Gram model for training, so as to obtain a vector of the type of the POI data, and determining the type of each city functional area according to the vector. The application can generate a walk sequence in a first-order neighborhood and / or a high-order neighborhood according to target POI data, input the walk sequence into a Skip-Gram model for training, so as to obtain a vector of each POI data type reflecting POI data homogeneity and structural similarity, and predict the type of each city functional area with high accuracy according to the vector and clustering analysis.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of city functional area recognition of POI data, and particularly relates to a city functional area recognition method considering POI homogeneity and structural similarity. BACKGROUND

[0002] With the advent of the big data era, scholars have gradually carried out in-depth research on the recognition of city functional areas. The recognition of city functional areas based on POI (Points of Interest) is one of the important directions. POI refers to the point data representing geographical entities on an electronic map, which has the characteristics of availability and completeness. POI not only contains the physical characteristics of geographical entities, but also represents the social and economic properties. This is in line with the characteristics of city functional areas, which represent both physical and social economic characteristics. At present, the word embedding model is an important method for recognizing city functional areas using POI data. This method aims to generate vector representations of POIs based on their spatial relationships, and then obtain vector representations of city functional areas for classification and recognition.

[0003] However, based on the spatial relationships of city functional area geographical elements, the method of extracting geographical element feature vectors only considers the homogeneity of geographical elements, and does not consider the structural similarity of geographical elements. This restricts the expression accuracy of geographical element features, reduces the feature expression accuracy of city functional areas, and further restricts the accuracy of city functional area recognition.

[0004] The above content is only used to assist in understanding the technical solutions of the present application and does not represent the acknowledgement of the above content as prior art. SUMMARY

[0005] The main purpose of the present application is to provide a city functional area recognition method, device, equipment and computer readable storage medium considering POI homogeneity and structural similarity, which aims to solve the technical problem that the existing method of extracting geographical element feature vectors only considers the homogeneity of geographical elements and does not consider the structural similarity of geographical elements.

[0006] To achieve the above purpose, the present application provides a city functional area recognition method considering POI homogeneity and structural similarity, which comprises the following steps:

[0007] Obtain each POI data in each city functional area, and determine the target POI data and the target city functional area where the target POI data is located.

[0008] For each target POI data, a plurality of circular regions with the target POI data as the center are determined according to a plurality of preset radii of different lengths, a buffer region intersected by the circular region and the target city functional region is determined, and a plurality of neighborhoods corresponding to the target POI data are determined according to the preset radii and the buffer region, wherein the neighborhoods include a first-order neighborhood and a plurality of high-order neighborhoods.

[0009] According to the breadth search algorithm, the depth search algorithm, and a preset walk length, the target POI data is sampled to generate a walk sequence corresponding to the target POI data in the first-order neighborhood and / or the high-order neighborhood.

[0010] For each city functional region, the walk sequence corresponding to the POI data in the city functional region is input into a Skip-Gram model for training to obtain a vector of each POI data type in each city functional region, and the type of each city functional region is determined according to the vector.

[0011] Further, the step of determining a plurality of neighborhoods according to the preset radii and the buffer region, wherein the neighborhoods include a first-order neighborhood and a plurality of high-order neighborhoods, comprises:

[0012] The number of the walk sequences is determined, and if the number does not reach a preset number of walk sequences, the step of sampling the target POI data to generate a walk sequence corresponding to the target POI data in the first-order neighborhood and / or the high-order neighborhood is returned.

[0013] Further, the step of sampling the target POI data according to the breadth search algorithm, the depth search algorithm, and a preset walk length to generate a walk sequence corresponding to the target POI data in the first-order neighborhood and / or the high-order neighborhood comprises:

[0014] According to the breadth search algorithm, the target POI data is sampled as a starting node to obtain a first node in the first-order neighborhood of the target POI data, and the first node is added to a target node.

[0015] A hyperparameter is configured to control the breadth search algorithm and the depth search algorithm to determine a random probability.

[0016] According to the random probability, a second node is obtained in the first-order neighborhood of the first node, the second node is added to the target node, and the number of target nodes is determined.

[0017] According to the number of target nodes and the walk length, a walk sequence corresponding to the target POI data is generated in the first-order neighborhood and / or the high-order neighborhood.

[0018] Further, the step of determining the walk sequence corresponding to the target POI data generated in the first-order neighborhood and / or the high-order neighborhood according to the number of target nodes and the walk length comprises:

[0019] If the number of target nodes does not reach the walk length, the second node is taken as the first node, and the step of obtaining the second node in the first-order neighborhood of the first node is performed again;

[0020] If the number of target nodes reaches the walk length, the type of POI data corresponding to the starting node and the target node is obtained, and the type is taken as the walk sequence.

[0021] Further, the step of determining a plurality of neighborhoods according to the preset radius and the buffer area, wherein the neighborhoods comprise a first-order neighborhood and a plurality of high-order neighborhoods comprises:

[0022] The preset radius is sorted, and according to the sorting result, a first target radius with the shortest radius length and a plurality of adjacent target radius groups are determined in each preset radius, wherein the target radius group comprises a second target radius and a third target radius, and the second target radius is smaller than the third target radius;

[0023] The first-order neighborhood is determined according to the first target radius, the buffer area and the urban function area, and the high-order neighborhood is determined according to the second target radius, the third target radius, the buffer area and the urban function area.

[0024] Further, the step of determining the first-order neighborhood according to the first target radius, the buffer area and the urban function area, and determining the high-order neighborhood according to the second target radius, the third target radius, the buffer area and the urban function area comprises:

[0025] A first buffer area corresponding to the first target radius in the buffer area is determined, and the first buffer area is taken as a first-order buffer area;

[0026] The first-order buffer area is superimposed with the target urban function area to obtain first POI data intersecting with the first-order buffer area, and the first POI data is taken as a first-order neighborhood;

[0027] A second buffer area corresponding to the second target radius and a third buffer area corresponding to the third target radius in the buffer area are determined, and an area in the third buffer area other than the second buffer area is taken as a high-order buffer area;

[0028] Superimpose the high-order buffer area and the target urban function area to obtain second POI data intersecting with the high-order buffer area, and take the second POI data as a high-order neighborhood.

[0029] Further, the step of determining the type of each urban function area according to the vector comprises:

[0030] For each urban function area, add the vectors in the urban function area to determine a target vector corresponding to the urban function area, and determine the number of vectors in the urban function area;

[0031] Divide the target vector by the number to obtain a feature vector corresponding to the urban function area, and perform cluster analysis on the feature vectors of the urban function areas to determine the type of each urban function area.

[0032] In addition, to achieve the above object, the application further provides a city function area identification device considering POI homogeneity and structural similarity, which comprises:

[0033] An acquisition module is configured to acquire each POI data in each urban function area, take the POI data in the same urban function area as target POI data, and determine a target urban function area where the target POI data is located;

[0034] A first determination module is configured to, for each target POI data, determine a plurality of circular areas with the target POI data as the center according to a plurality of preset radii of different lengths, determine a buffer area intersecting the target urban function area, and determine a plurality of neighborhoods corresponding to the target POI data according to the preset radii and the buffer area, wherein the neighborhoods comprise a first-order neighborhood and a plurality of high-order neighborhoods.

[0035] A second determination module is configured to sample the target POI data according to a breadth-first search algorithm, a depth-first search algorithm, and a preset walk length, and generate a walk sequence corresponding to the target POI data in the first-order neighborhood and / or the high-order neighborhood.

[0036] A third determination module is configured to, for each urban function area, input the walk sequence corresponding to the POI data in the urban function area into a Skip-Gram model for training to obtain a type vector of the POI data in each urban function area, and determine the type of each urban function area according to the vector.

[0037] In addition, to achieve the above object, the present application also provides a city functional area identification device considering POI homogeneity and structural similarity, which comprises a memory, a processor and a city functional area identification program considering POI homogeneity and structural similarity stored on the memory and capable of running on the processor, and the city functional area identification program considering POI homogeneity and structural similarity realizes the steps of the aforementioned city functional area identification method considering POI homogeneity and structural similarity when executed by the processor.

[0038] In addition, to achieve the above object, the present application also provides a computer readable storage medium, which stores a city functional area identification program considering POI homogeneity and structural similarity, and the city functional area identification program considering POI homogeneity and structural similarity realizes the steps of the aforementioned city functional area identification method considering POI homogeneity and structural similarity when executed by a processor.

[0039] The present application obtains each POI data in each preset city functional area, takes the POI data in the same city functional area as target POI data, and determines the target city functional area where the target POI data is located, then for each target POI data, determines a plurality of circular areas with the target POI data as the center according to a plurality of preset radii of different lengths, determines a buffer area intersected by the circular area and the target city functional area, and determines a plurality of neighborhoods corresponding to the target POI data according to the preset radius and the buffer area, wherein the neighborhood comprises a first-order neighborhood and a plurality of high-order neighborhoods, then samples the target POI data according to the breadth search algorithm, the depth search algorithm and a preset walk length, generates a walk sequence corresponding to the target POI data in the first-order neighborhood and / or the high-order neighborhood, then inputs the walk sequence corresponding to the POI data in each city functional area into a Skip-Gram model for training to obtain a vector of each POI data type in each city functional area, and determines the type of each city functional area according to the vector, which can generate a walk sequence in the first-order neighborhood and / or the high-order neighborhood according to the target POI data, and after inputting into the Skip-Gram model for training, obtains a vector of each POI data type reflecting POI data homogeneity and structural similarity, and according to the vector and clustering analysis, predicts the type of each city functional area with high accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0040] Figure 1 is a structural schematic diagram of a city functional area identification device considering POI homogeneity and structural similarity in a hardware running environment involved in the embodiment scheme of the present application.

[0041] Figure 2 This is a flowchart illustrating the first embodiment of the urban functional area identification method of the present invention, which takes into account both POI homogeneity and structural similarity.

[0042] Figure 3 This is a schematic diagram of the functional modules of an embodiment of the urban functional area identification device of the present invention, which takes into account both the homogeneity and structural similarity of POIs.

[0043] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0044] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0045] like Figure 1 As shown, Figure 1 This is a schematic diagram of the structure of an urban functional area identification device that balances POI homogeneity and structural similarity in the hardware operating environment involved in the embodiments of the present invention.

[0046] The urban functional area identification device of the present invention, which takes into account both POI homogeneity and structural similarity, can be a PC, or a smartphone, tablet computer, e-book reader, MP3 (Moving Picture Experts Group Audio Layer III) player, MP4 (Moving Picture Experts Group Audio Layer IV) player, portable computer, or other mobile terminal device with display function.

[0047] like Figure 1 As shown, the urban functional area identification device that balances POI homogeneity and structural similarity may include: a processor 1001, such as a CPU; a network interface 1004; a user interface 1003; a memory 1005; and a communication bus 1002. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen and an input unit such as a keyboard; optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The memory 1005 may be high-speed RAM or stable non-volatile memory, such as a disk drive. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0048] Optionally, the city functional area identification device considering POI homogeneity and structural similarity can also include a camera, an RF (Radio Frequency) circuit, a sensor, an audio circuit, a WiFi module, and the like. The sensor can be, for example, a light sensor, a motion sensor, and other sensors. Specifically, the light sensor can include an ambient light sensor and a proximity sensor. The ambient light sensor can adjust the brightness of the display screen according to the brightness of the ambient light, and the proximity sensor can turn off the display screen and / or backlight when the city functional area identification device considering POI homogeneity and structural similarity is moved to the ear. As one of the motion sensors, the gravity acceleration sensor can detect the size of acceleration in each direction (generally three axes), and when at rest, it can detect the size and direction of gravity, which can be used for applications such as identifying the posture of the city functional area identification device considering POI homogeneity and structural similarity (such as landscape / portrait switching, related games, magnetometer posture calibration), vibration recognition related functions (such as pedometer, knocking), and the like. Of course, the city functional area identification device considering POI homogeneity and structural similarity can also be configured with a gyroscope, a barometer, a hygrometer, a thermometer, an infrared sensor, and other sensors, which will not be described here.

[0049] Those skilled in the art can understand that Figure 1 The terminal structure shown in the figure does not constitute a limitation on the terminal, and can include more or fewer components than the figure, or combine certain components, or different component arrangements.

[0050] As shown in Figure 1 As a computer storage medium, the memory 1005 can include an operating system, a network communication module, a user interface module, and a city functional area identification program considering POI homogeneity and structural similarity.

[0051] In the terminal shown in Figure 1 The network interface 1004 is mainly used to connect to a background server and communicate data with the background server. The user interface 1003 is mainly used to connect to a client (user end) and communicate data with the client. The processor 1001 can be used to call the city functional area identification program considering POI homogeneity and structural similarity stored in the memory 1005.

[0052] In the embodiment, the city function area identification device considering POI homogeneity and structural similarity comprises a memory 1005, a processor 1001, and a city function area identification program considering POI homogeneity and structural similarity stored in the memory 1005 and capable of running on the processor 1001. When the processor 1001 invokes the city function area identification program considering POI homogeneity and structural similarity stored in the memory 1005, the steps of the city function area identification method considering POI homogeneity and structural similarity in each embodiment are performed.

[0053] The application further provides a city function area identification method considering POI homogeneity and structural similarity, which refers to Figure 2 , Figure 2 The flowchart of the first embodiment of the method of the application is shown in FIG. 1.

[0054] In the embodiment, the city function area identification method considering POI homogeneity and structural similarity comprises the following steps.

[0055] In step S101, each POI data in each city function area is obtained, the POI data in the same city function area is taken as target POI data, and a target city function area where the target POI data is located is determined.

[0056] In the embodiment, the POI point data on the map can be obtained by using GPS software, and each city function area is obtained by manually dividing the area on the map. The POI data in the same city function area is taken as target POI data, and a target city area where the target POI data is located is determined as the target city area.

[0057] In step S102, for each target POI data, a plurality of circular areas with the target POI data as the center are determined according to a plurality of preset radii with different lengths, a buffer area where the circular area intersects with the target city function area is determined, and a plurality of neighborhoods corresponding to the target POI data are determined according to the preset radii and the buffer area, wherein the neighborhoods comprise a first-order neighborhood and a plurality of high-order neighborhoods.

[0058] In the embodiment, for each target POI data, the target POI data is taken as the center, and a plurality of preset radii with different lengths, for example, 3 cm and 4 cm, are used to determine a plurality of circular areas with the target POI data as the center. The area where the circular area intersects with the city function area is taken as the buffer area, wherein the buffer area is in the target city function area where the target POI data is located. Then, according to the preset radii and the buffer area, a plurality of neighborhoods corresponding to the target POI data are determined in the target city function area, wherein the neighborhoods comprise a first-order neighborhood and a plurality of high-order neighborhoods.

[0059] Further, in an embodiment, step S102 further comprises:

[0060] a. sorting the preset radii, and determining, according to a result of the sorting, a first target radius with a shortest length among the preset radii, and a plurality of target radius groups each including a second target radius and a third target radius, the second target radius being smaller than the third target radius;

[0061] b. determining a first-order neighborhood according to the first target radius, a buffer region, and the urban functional region, and determining a high-order neighborhood according to the second target radius, the third target radius, the buffer region, and the urban functional region.

[0062] In the embodiment, for each target POI data, the preset radii are sorted, and according to a result of the sorting, a first target radius with a shortest length among the preset radii is determined, and a plurality of target radius groups each including a second target radius and a third target radius are determined, the second target radius being smaller than the third target radius, then a first-order neighborhood is determined according to the first target radius, a buffer region, and the urban functional region, and a high-order neighborhood is determined according to the second target radius, the third target radius, the buffer region, and the urban functional region.

[0063] For example, according to a result of the sorting, the preset radii are 3 cm, 4 cm, and 5 cm, the 3 cm radius is determined as the first target radius, three buffer regions are determined according to the 3 cm radius, the 4 cm radius, and the 5 cm radius respectively, the 3 cm radius and the 4 cm radius are determined as a target radius group, and the 4 cm radius and the 5 cm radius are determined as another target radius group.

[0064] Further, in an embodiment, step b further comprises:

[0065] b1. determining a first buffer region corresponding to the first target radius in the buffer region, and taking the first buffer region as a first-order buffer region;

[0066] b2. superimposing the first-order buffer region and the target urban functional region to obtain first POI data intersecting with the first-order buffer region, and taking the first POI data as a first-order neighborhood;

[0067] b3. determining a second buffer region corresponding to the second target radius and a third buffer region corresponding to the third target radius in the buffer region, and taking a region in the third buffer region other than the second buffer region as a high-order buffer region;

[0068] Step b4, superimposing the high-order buffer area and the target urban function area to obtain second POI data intersecting with the high-order buffer area, and taking the second POI data as a high-order neighborhood.

[0069] In this embodiment, a first buffer area corresponding to a first target radius is determined in the buffer area, and the first buffer area is taken as a first-order buffer area. POI data in the first-order buffer area is superimposed with POI data in the target urban function area to obtain first POI data intersecting with the target urban function area and the first-order buffer area, and the first POI data is taken as a first-order neighborhood.

[0070] In this embodiment, a first buffer area corresponding to a first target radius is determined in the buffer area, and the first buffer area is taken as a first-order buffer area. POI data in the first-order buffer area is superimposed with POI data in the target urban function area to obtain first POI data intersecting with the target urban function area and the first-order buffer area, and the first POI data is taken as a first-order neighborhood.

[0071] Step S103, according to the breadth search algorithm, the depth search algorithm, and the preset walk length, sampling the target POI data to generate a walk sequence corresponding to the target POI data in the first-order neighborhood and / or the high-order neighborhood.

[0072] In this embodiment, the target POI data can be sampled according to the breadth search algorithm, the depth search algorithm, and the preset walk length, with the target POI data as a sampling starting point, to generate a walk sequence in the first-order neighborhood and the high-order neighborhood.

[0073] Further, in an embodiment, step S103 further includes:

[0074] Step c, if the number of the walk sequences does not reach the preset number of walk sequences, returning to the step of sampling the target POI data to generate a walk sequence corresponding to the target POI data in the first-order neighborhood and / or the high-order neighborhood.

[0075] In the embodiment, the number of the walk sequences corresponding to the target POI data is determined by setting the number of the walk sequences artificially. If the number of the walk sequences of the target POI data does not reach the number of the walk sequences set artificially, the target POI data is sampled to generate the walk sequences corresponding to the target POI data in the first-order neighborhood and / or the high-order neighborhood. If the number of the walk sequences corresponding to the target POI data reaches the number of the walk sequences set artificially, the walk sequences corresponding to the POI data in the same city function area are input into the Skip-Gram model to be trained to obtain the vector of the type of the POI data, and the type of each city function area is determined according to the vector.

[0076] Further, in an embodiment, step S103 further includes:

[0077] Step d, according to the breadth search algorithm, the target POI data is sampled as a starting node to obtain a first node in the first-order neighborhood of the target POI data, and the first node is added to the target node;

[0078] Step e, configuring a hyperparameter to control the breadth search algorithm and the depth search algorithm to determine a random probability;

[0079] Step f, according to the random probability, a second node is obtained in the first-order neighborhood of the first node, the second node is added to the target node, and the number of target nodes is determined;

[0080] Step g, according to the number of target nodes and the walk length, the walk sequences corresponding to the target POI data are generated in the first-order neighborhood and / or the high-order neighborhood.

[0081] In the embodiment, according to the breadth search algorithm, the target POI data is sampled as a starting node to obtain a first node in the first-order neighborhood of the target POI data, and the first node is added to the target node. Then, by configuring a hyperparameter to control the breadth search algorithm and the depth search algorithm, a random probability is determined, according to the random probability, a second node is obtained in the first-order neighborhood of the first node. It should be noted that the first node and the second node are both in the target city function area where the target POI data is located. Then, the second node is added to the target node, and the number of target nodes is determined. According to whether the number of target nodes reaches the preset walk length, it is determined whether the walk sequences corresponding to the target POI data are generated in the first-order neighborhood and / or the high-order neighborhood of the target POI data.

[0082] Further, in an embodiment, step h further includes:

[0083] Step h1, if the number of target nodes does not reach the length of the walk, the second node is taken as the first node, and the step of acquiring the second node in the first-order neighborhood of the first node is executed again.

[0084] Step h2, if the number of target nodes reaches the length of the walk, the type of POI data corresponding to the starting node and the target node is acquired, and the type is taken as the walk sequence.

[0085] In this embodiment, first, it is determined whether the number of target nodes reaches the length of the walk, if the number of target nodes does not reach the length of the walk, the second node is taken as the first node, and the step of acquiring the second node in the first-order neighborhood of the first node is executed again, and the loop is repeated until it is determined that the number of target nodes reaches the length of the walk.

[0086] Step h2, if the number of target nodes reaches the length of the walk, the type of POI data corresponding to the starting node and the target node is acquired, and the type is taken as the walk sequence.

[0087] In this embodiment, if the number of target nodes reaches the length of the walk, the type of POI data corresponding to the starting node and the target node is acquired, and the type of POI data is taken as the walk sequence, wherein the POI data includes type data, for example, the type data can be a small shop, a supermarket, a bank, a hospital, a school, etc.

[0088] Step S104, for each urban functional area, the walk sequence corresponding to the POI data in the urban functional area is input into the Skip-Gram model for training to obtain the vector of each POI data type in each urban functional area, and the type of each urban functional area is determined according to the vector.

[0089] In this embodiment, for each urban functional area, the walk sequence corresponding to the POI data in the urban functional area is input into the Skip-Gram model for training to obtain the vector of each POI data type in each urban functional area, the similarity between these POI type vectors can reflect the homogeneity and structural similarity of POI, then the type of each urban functional area is determined according to the vector, wherein the homogeneity refers to that the POI with similar spatial distance has a higher similarity of vectors, and the structural similarity refers to that the POI with similar neighborhood structure has a higher similarity of vectors.

[0090] For example, the sequence is input into a word vector model for model training to obtain a vector, and the vector is processed to obtain a feature vector of the urban functional area. Then, a clustering method such as a K-means algorithm is used to process the feature vector, and the type of each urban functional area is determined. Commonly used word embedding training models include CBOW and Skip-Gram, and the training result of Skip-Gram has higher precision. Therefore, the Skip-Gram model can be selected for training. The Skip-Gram model is opposite to the CBOW model. The CBOW model predicts the current value through the context, and the Skip-Gram model predicts the context through the current value.

[0091] It should be noted that the K-means algorithm is an unsupervised clustering algorithm based on distance grouping. For a given sample set, the samples are divided into K clusters according to the distance between the samples, so that the points in the cluster are closely connected, and the distance between the points in different clusters is as large as possible. The processing process of the K-means algorithm includes the following steps: step 1: randomly select K points as clustering centers, that is, k class center vectors; step 2: calculate the distance from each sample point to each class center vector, and divide it into the nearest class; step 3: update the center vector of each class; step 4: determine whether the new class center vector changes, if it changes, go to step 2, if the class center vector no longer changes, stop and output the clustering result.

[0092] The city function area recognition method considering POI homogeneity and structural similarity provided in the embodiment obtains preset POI data in each city function area, takes POI data in the same city function area as target POI data, and determines a target city function area where the target POI data is located. Then, for each target POI data, a plurality of circular regions with the target POI data as the center are determined according to a plurality of preset radii of different lengths, a buffer region intersecting the circular region and the target city function area is determined, and a plurality of neighborhoods corresponding to the target POI data are determined according to the preset radii and the buffer region. The neighborhoods include a first-order neighborhood and a plurality of high-order neighborhoods. Then, the target POI data is sampled according to the breadth search algorithm, the depth search algorithm, and a preset walk length, a walk sequence corresponding to the target POI data is generated in the first-order neighborhood and / or the high-order neighborhood, and then the walk sequence corresponding to the POI data in each city function area is input into a Skip-Gram model for training to obtain a vector of each POI data type in each city function area. The type of each city function area is determined according to the vector. The walk sequence in the first-order neighborhood and / or the high-order neighborhood can be generated according to the target POI data, and the vector of each POI data type reflecting POI homogeneity and structural similarity can be obtained after the walk sequence is input into the Skip-Gram model for training. The type of each city function area can be predicted with high accuracy according to the vector and clustering analysis.

[0093] Based on the first embodiment, a second embodiment of the city function area recognition method considering POI homogeneity and structural similarity is provided. In the second embodiment, step S104 includes:

[0094] In step S201, for each city function area, the vectors in the city function area are added to determine a target vector corresponding to the city function area, and the number of vectors in the city function area is determined.

[0095] In step S202, the target vector is divided by the number to obtain a feature vector corresponding to the city function area, and clustering analysis is performed on the feature vectors of each city function area to determine the type of each city function area.

[0096] In the embodiment, for each urban functional area, vectors in the urban functional area are added to obtain a first calculation result, the first calculation result is taken as a target vector corresponding to the urban functional area, the number of vectors in the urban functional area is determined, and the target vector is divided by the number to obtain a second calculation result, the second calculation result is taken as a feature vector corresponding to the urban functional area, and finally, clustering analysis is performed on the feature vectors of the urban functional areas to determine the type of each urban functional area.

[0097] The urban functional area recognition method considering POI homogeneity and structural similarity provided in the embodiment can determine the target vector corresponding to each urban functional area by adding the vectors in the urban functional area, determine the number of vectors in the urban functional area, then divide the target vector by the number to obtain the feature vector corresponding to the urban functional area, and perform clustering analysis on the feature vectors of the urban functional areas to determine the type of each urban functional area, so that each POI data type vector reflecting POI data homogeneity and structural similarity can be used to predict the type of each urban functional area with high accuracy.

[0098] The application further provides a device for recognizing urban functional areas considering POI homogeneity and structural similarity. Figure 3 The device for recognizing urban functional areas considering POI homogeneity and structural similarity comprises:

[0099] The acquisition module 10 acquires each POI data in each urban functional area, takes the POI data in the same urban functional area as target POI data, and determines the target urban functional area where the target POI data is located.

[0100] The first determination module 20 is configured to determine, for each target POI data, a plurality of circular regions with the target POI data as the center according to a plurality of preset radii of different lengths, determine a buffer region intersected by the circular regions and the target urban functional area, and determine a plurality of neighborhoods corresponding to the target POI data according to the preset radii and the buffer region, wherein the neighborhoods include a first-order neighborhood and a plurality of high-order neighborhoods.

[0101] The second determination module 30 is configured to sample the target POI data according to a breadth-first search algorithm, a depth-first search algorithm and a preset walk length, and generate a walk sequence corresponding to the target POI data in the first-order neighborhood and / or the high-order neighborhood.

[0102] The third determining module 40 is configured to input the walk sequence corresponding to the POI data in each urban functional region into a Skip-Gram model for training, so as to obtain a vector of each POI data type in each urban functional region, and determine the type of each urban functional region according to the vector.

[0103] Further, the second determining module 30 is configured to:

[0104] determine the number of the walk sequences, and if the number does not reach a preset number of walk sequences, return to perform sampling on the target POI data to generate the walk sequence corresponding to the target POI data in the first-order neighborhood and / or the high-order neighborhood.

[0105] Further, the second determining module 30 is configured to:

[0106] According to the breadth search algorithm, sample the target POI data as a starting node to obtain a first node in the first-order neighborhood of the target POI data, and add the first node to the target node.

[0107] configure a hyperparameter to control the breadth search algorithm and the depth search algorithm to determine a random probability.

[0108] According to the random probability, obtain a second node in the first-order neighborhood of the first node, add the second node to the target node, and determine the number of the target nodes.

[0109] According to the number of the target nodes and the walk length, determine the walk sequence corresponding to the target POI data in the first-order neighborhood and / or the high-order neighborhood.

[0110] Further, the second determining module 30 is configured to:

[0111] If the number of the target nodes does not reach the walk length, take the second node as the first node, and return to perform the obtaining of the second node in the first-order neighborhood of the first node.

[0112] If the number of the target nodes reaches the walk length, obtain the type of the POI data corresponding to the starting node and the target node, and take the type as the walk sequence.

[0113] Further, the first determining module 20 is configured to:

[0114] sort the preset radii, according to the sorting result, determine a first target radius with the shortest radius length and a plurality of groups of adjacent target radius groups in the preset radii, wherein the target radius group includes a second target radius and a third target radius, and the second target radius is smaller than the third target radius.

[0115] According to the first target radius, the buffer area and the urban functional area, a first-order neighborhood is determined, and according to the second target radius, the third target radius, the buffer area and the urban functional area, a high-order neighborhood is determined.

[0116] Further, the first determining module 20 is configured to:

[0117] determine a first buffer area corresponding to the first target radius in the buffer area, and take the first buffer area as a first-order buffer area;

[0118] superimpose the first-order buffer area and the target urban functional area to obtain first POI data intersecting with the first-order buffer area, and take the first POI data as a first-order neighborhood;

[0119] determine a second buffer area corresponding to the second target radius and a third buffer area corresponding to the third target radius in the buffer area, and take an area in the third buffer area except the second buffer area as a high-order buffer area;

[0120] superimpose the high-order buffer area and the target urban functional area to obtain second POI data intersecting with the high-order buffer area, and take the second POI data as a high-order neighborhood.

[0121] Further, the third determining module 40 is further configured to:

[0122] add vectors in each urban functional area to determine a target vector corresponding to the urban functional area, and determine a number of vectors in the urban functional area;

[0123] divide the target vector by the number to obtain a feature vector corresponding to the urban functional area, and perform clustering analysis on the feature vectors of the urban functional areas to determine types of the urban functional areas.

[0124] The method performed by each program unit can refer to each embodiment of the urban functional area recognition method taking into account POI homogeneity and structural similarity of the present application, which will not be described here.

[0125] In addition, the embodiment of the present application also provides a city function area identification device considering POI homogeneity and structural similarity, which comprises a memory, a processor and a city function area identification program considering POI homogeneity and structural similarity stored in the memory and capable of running on the processor, and the city function area identification program considering POI homogeneity and structural similarity realizes the steps of the city function area identification method considering POI homogeneity and structural similarity when executed by the processor.

[0126] In addition, the embodiment of the present application also provides a computer readable storage medium, which stores a city function area identification program considering POI homogeneity and structural similarity, and the city function area identification program considering POI homogeneity and structural similarity realizes the steps of the city function area identification method considering POI homogeneity and structural similarity when executed by a processor.

[0127] It should be noted that, in this document, the terms "comprising", "containing", or any other similar term are intended to encompass non-exclusive inclusion, so that a process, method, article or system that comprises a list of elements does not only include those elements, but also includes other elements not explicitly listed, or further includes elements inherent in such process, method, article or system. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of additional identical elements in the process, method, article or system comprising the element.

[0128] The above-mentioned embodiment numbers of the present application are only for description, and do not represent the advantages or disadvantages of the embodiments.

[0129] Through the above description of the embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be realized by means of software and necessary general hardware platforms, of course, they can also be realized by hardware, but in many cases, the former is a better embodiment. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which is stored in a storage medium (such as a ROM / RAM, a magnetic disk, an optical disk) and includes a plurality of instructions for making a terminal device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device) execute the methods described in the embodiments of the present application.

[0130] The above is only the preferred embodiment of the present application, and does not limit the patent scope of the present application, and any equivalent structure or equivalent process transformation made by using the content of the specification and drawings, or direct or indirect application in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for identifying urban functional areas that balances POI homogeneity and structural similarity, characterized in that, The urban functional area identification method that takes into account both POI homogeneity and structural similarity includes the following steps: Acquire POI data for each preset city functional area, use POI data within the same city functional area as target POI data, and determine the target city functional area where the target POI data is located. For each target POI data, multiple circular regions centered on the target POI data are determined based on multiple preset radii of different lengths. A buffer region is determined where the circular regions intersect with the target urban functional area. Based on the preset radii and the buffer region, multiple neighborhoods corresponding to the target POI data are determined, wherein the neighborhoods include first-order neighborhoods and multiple higher-order neighborhoods. Based on the breadth-first search algorithm, the depth-first search algorithm, and the preset walk length, the target POI data is sampled, and the walk sequence corresponding to the target POI data is generated in the first-order neighborhood and / or the higher-order neighborhood. For each urban functional area, the walk sequence corresponding to the POI data within the urban functional area is input into the Skip-Gram model for training to obtain the vector of each POI data type within each urban functional area, and the type of each urban functional area is determined based on the vector. The step of sampling the target POI data according to the breadth-first search algorithm, the depth-first search algorithm, and the preset walk length, and generating the walk sequence corresponding to the target POI data in the first-order neighborhood and / or the higher-order neighborhood includes: According to the breadth-first search algorithm, the target POI data is sampled as the starting node to obtain the first node in the first-order neighborhood of the target POI data, and the first node is added to the target node. The intensity of the breadth-first search algorithm and the depth-first search algorithm is controlled by configuring hyperparameters to determine the random probability; Based on the random probability, a second node is obtained within the first-order neighborhood of the first node, the second node is added to the target node, and the number of target nodes is determined. Based on the number of target nodes and the walk length, determine the walk sequence corresponding to generating target POI data within the first-order neighborhood and / or the higher-order neighborhood.

2. The urban functional area identification method that balances POI homogeneity and structural similarity as described in claim 1, characterized in that, The step of determining multiple neighborhoods based on the preset radius and the buffer region, wherein the neighborhoods include first-order neighborhoods and multiple higher-order neighborhoods, includes: The number of walk sequences is determined. If the number does not reach the preset number of walk sequences, the process returns to the step of sampling the target POI data to generate walk sequences corresponding to the target POI data in the first-order neighborhood and / or the higher-order neighborhood.

3. The urban functional area identification method that balances POI homogeneity and structural similarity as described in claim 1, characterized in that, The step of determining the walk sequence corresponding to the target POI data generated in the first-order neighborhood and / or the higher-order neighborhood based on the number of target nodes and the walk length includes: If the number of target nodes does not reach the specified walking length, the second node is treated as the first node, and the process returns to retrieve the second node within the first-order neighborhood of the first node. If the number of target nodes reaches the specified walk length, then the type of POI data corresponding to the starting node and the target node is obtained, and the type is used as the walk sequence.

4. The urban functional area identification method that takes into account both POI homogeneity and structural similarity as described in claim 1, characterized in that, The step of determining multiple neighborhoods based on the preset radius and the buffer region, wherein the neighborhoods include first-order neighborhoods and multiple higher-order neighborhoods, includes: The preset radii are sorted, and based on the sorting result, a first target radius with the shortest radius length and multiple groups of adjacent target radii are determined among the preset radii. The target radius group includes a second target radius and a third target radius, wherein the second target radius is smaller than the third target radius. A first-order neighborhood is determined based on the first target radius, the buffer zone, and the urban functional area. A higher-order neighborhood is determined based on the second target radius, the third target radius, the buffer zone, and the urban functional area.

5. The urban functional area identification method that takes into account both POI homogeneity and structural similarity as described in claim 4, characterized in that, The steps of determining a first-order neighborhood based on the first target radius, the buffer zone, and the urban functional area, and determining a higher-order neighborhood based on the second target radius, the third target radius, the buffer zone, and the urban functional area, include: In the buffer region, a first buffer region corresponding to the first target radius is determined, and the first buffer region is regarded as a first-order buffer region. The first-order buffer area is superimposed with the target city functional area to obtain the first POI data that intersects with the first-order buffer, and the first POI data is used as the first-order neighborhood. In the buffer region, a second buffer region corresponding to the second target radius and a third buffer region corresponding to the third target radius are determined, and the region in the third buffer region other than the second buffer region is regarded as a higher-order buffer region. The higher-order buffer area is superimposed on the target city functional area to obtain second POI data that intersects with the higher-order buffer, and the second POI data is used as the higher-order neighborhood.

6. The urban functional area identification method that takes into account both POI homogeneity and structural similarity as described in claim 1, characterized in that, The step of determining the type of each urban functional area based on the vector includes: For each urban functional area, the vectors within the urban functional area are summed to determine the target vector corresponding to the urban functional area and to determine the number of vectors within the urban functional area. Divide the target vector by the number to obtain the feature vector corresponding to the urban functional area, and perform cluster analysis on the feature vector of each urban functional area to determine the type of each urban functional area.

7. A device for identifying urban functional areas that balances POI homogeneity and structural similarity, characterized in that, The urban functional area identification device that takes into account both POI homogeneity and structural similarity includes: The acquisition module acquires POI data from each preset city functional area, uses POI data from the same city functional area as target POI data, and determines the target city functional area where the target POI data is located. The first determining module is used to determine, for each target POI data, multiple circular regions centered on the target POI data according to multiple preset radii of different lengths, determine a buffer region where the circular regions intersect with the target urban functional area, and determine multiple neighborhoods corresponding to the target POI data according to the preset radii and the buffer regions, wherein the neighborhoods include first-order neighborhoods and multiple higher-order neighborhoods. The second determining module is used to sample the target POI data according to a breadth-first search algorithm, a depth-first search algorithm, and a preset walk length, and generate a walk sequence corresponding to the target POI data within the first-order neighborhood and / or the higher-order neighborhood. The second determining module is also used to sample the target POI data as a starting node according to the breadth-first search algorithm to obtain a first node within the first-order neighborhood of the target POI data, and add the first node to the target node; control the intensity of the breadth-first search algorithm and the depth-first search algorithm by configuring hyperparameters to determine the random probability; obtain a second node within the first-order neighborhood of the first node according to the random probability, add the second node to the target node, and determine the number of target nodes; and determine the walk sequence corresponding to the target POI data generated within the first-order neighborhood and / or the higher-order neighborhood based on the number of target nodes and the walk length. The third determination module is used to input the walk sequence corresponding to the POI data in each urban functional area into the Skip-Gram model for training, so as to obtain the vector of each POI data type in each urban functional area, and determine the type of each urban functional area based on the vector.

8. A device for identifying urban functional areas that balances POI homogeneity and structural similarity, characterized in that, The urban functional area identification device that takes into account both POI homogeneity and structural similarity includes: a memory, a processor, and an urban functional area identification program that takes into account both POI homogeneity and structural similarity, stored in the memory and executable on the processor. When the urban functional area identification program that takes into account both POI homogeneity and structural similarity is executed by the processor, it implements the steps of the urban functional area identification method that takes into account both POI homogeneity and structural similarity as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a city functional area identification program that takes into account both POI homogeneity and structural similarity. When the city functional area identification program that takes into account both POI homogeneity and structural similarity is executed by a processor, it implements the steps of the city functional area identification method that takes into account both POI homogeneity and structural similarity as described in any one of claims 1 to 6.

Images