Personalized interest area extraction method and system based on interest unit

By extracting urban road data and sign-in data, dynamically adjusting the convolution layer weights, and combining with the improvement of clustering algorithm, the problem of fuzzy boundaries of interest regions in the existing technology is solved, and the precise and personalized extraction of interest regions is achieved.

CN119091116AActive Publication Date: 2024-12-06CHINESE ACAD OF SURVEYING & MAPPING
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202411094118.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-09
Publication Date
2024-12-06
Estimated Expiration
2044-08-09

AI Technical Summary

Technical Problem

When acquiring user interest areas (ROIs), the prior art lacks sufficient consideration of the actual geographical scope, resulting in the ROI boundary may be too large or too small, and there is ambiguity.

Method used

The personalized interest area extraction method based on interest units is divided by extracting urban road data, dynamically allocating weight parameters to the convolutional layer, extracting spatial and temporal fusion features, and clustering through improved clustering algorithms to accurately divide interest areas.

Benefits of technology

Accurate extraction of the boundaries of the interest zone is achieved, which reduces ambiguity and improves the accuracy and personalization of the interest zone division.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119091116A_ABST
    Figure CN119091116A_ABST
Patent Text Reader

Abstract

The embodiment of the invention provides a personalized interest area extraction method and system based on interest units, and the method comprises the steps: extracting urban road data in a research area, and dividing the research area into different interest units based on the urban road data; dynamically distributing corresponding weight parameters for the convolutional layer based on the sign-in data of the interest unit, extracting the time interval feature and the spatial distance feature of the interest unit, and fusing the above features to form a space-time fusion feature; and taking the space-time fusion features as input, clustering interest units through an improved clustering algorithm, dividing different interest areas, and realizing extraction of personalized interest areas.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This document relates to the field of urban planning technology, and in particular to a method and system for extracting personalized interest areas based on interest units. Background Art

[0002] At present, the traditional methods for obtaining user Region of Interest (ROI) mainly use grid or clustering methods. However, these methods have a common problem, that is, they lack sufficient consideration of the actual geographical scope, so the ROI boundary may not be accurate enough, may be too large or too small, and the final ROI boundary has a certain degree of ambiguity. In order to solve this problem, a personalized ROI extraction method based on interest units is proposed. Summary of the invention

[0003] One or more embodiments of this specification provide a method for extracting a personalized interest region based on an interest unit, including:

[0004] Extracting urban road data within a study area, and dividing the study area into different units of interest based on the urban road data;

[0005] Based on the check-in data of the interest unit, dynamically assign corresponding weight parameters to the convolution layer, extract the time interval feature and spatial distance feature of the interest unit, and fuse the above features to form a spatiotemporal fusion feature;

[0006] The spatiotemporal fusion features are used as input, and the interest units are clustered through an improved clustering algorithm to divide different interest areas, thereby realizing the extraction of personalized interest areas.

[0007] Furthermore, the urban road data includes data on expressways, main roads and urban main roads in OpenStreetMap;

[0008] Extracting urban road data within the study area, and dividing the study area into different interest units based on the urban road data specifically includes:

[0009] For the study area, the highway, main road and urban trunk road data in OpenStreetMap were selected as urban road data;

[0010] Preprocess the selected urban road data to form new road data. The specific steps include:

[0011] When the new road data has topological errors and thus generates irregular grids, the topological errors are modified so that the new road data forms closed units, and finally forms interest units based on the urban road data.

[0012] Furthermore, the screened urban road data is preprocessed to form new road data, specifically including:

[0013] Remove duplicate, hanging and independent roads, and extend and complete disconnected roads;

[0014] Generate buffer zones of different ranges for roads of different levels;

[0015] Based on urban road data and administrative district data, the noise and redundant information of urban road data are removed, the overlapping parts of road and administrative district data are integrated through overlay analysis, and the overlapping areas of administrative district data are cleaned.

[0016] Furthermore, based on the check-in data of the interest unit, the corresponding weight parameters are dynamically assigned to the convolution layer to extract the time interval features and spatial distance features of the interest unit, specifically including:

[0017] Set the spatial graph G = {V, E, A}, where nodes V = {v0, v1, v2...v n-1} is the interest unit, and the edge set E = {e0,e1,e3,…e m-1}, each e i represents the edge connecting two nodes, and the adjacency matrix A represents the spatial relationship between nodes. ij Indicates the unit of interest v i and v j Haversine distance between them;

[0018] Based on the check-in data of interest units, the check-in frequency of different interest units is analyzed, and the activity of each interest unit is calculated according to the check-in frequency; according to the activity of each interest unit, different convolution weights are dynamically assigned to the convolution layer, and then applied to multi-layer graph convolution and dilated convolution networks to capture the complex spatiotemporal relationship between interest units;

[0019] After normalizing the adjacency matrix, a graph convolution operation is performed, and the convolution weights are introduced into each graph convolution layer. Multiple layers of graph convolution are stacked to increase the network depth and extract the spatial distance features of the units of interest.

[0020] The dilated convolutional network is introduced to expand the convolution receptive field by introducing a dilation factor in the time dimension. On this basis, a dynamic convolution mechanism is further introduced to dynamically adjust the convolution kernel weights to adapt to the feature changes at different time scales and extract the time interval features of the units of interest.

[0021] Furthermore, the specific method of fusing the spatial distance feature and the time interval feature to form the spatiotemporal fusion feature is as follows:

[0022] In the same dimensional space, a mapping relationship between spatial interval features and time interval features is established, and the spatial interval features and time interval features are fused in a homomorphic manner to obtain the spatiotemporal fusion feature f m .

[0023] Furthermore, the spatiotemporal fusion features are used as input, and the interest units are clustered by an improved clustering algorithm to divide different interest areas, so as to realize the extraction of personalized interest areas. The specific method is as follows:

[0024] A particle swarm algorithm is designed to optimize cluster centers and unsupervised classification is used to divide similar units of interest into the same area.

[0025] The specific method of designing particle swarm algorithm to optimize cluster centers is as follows:

[0026] First, k particles are randomly generated, each particle contains the location of a possible initial cluster center;

[0027] For each particle, the fitness is calculated using the objective function of K-means clustering, where the objective function is to maximize the sum of the distance differences between the center cluster and the center of the cluster to which it belongs;

[0028] Calculate the particle cluster center position and moving direction to update the particle position. The specific calculation method is as follows:

[0029] x i (t+1)=x i (t)+v i (t+1);

[0030] v i (t+1)=w·v i (t)+c1·r1·(pbest i -x i (t))+c2·r2·(gbest-x i (t));

[0031] where x i (t) and v i (t) represent the particle cluster center position and moving direction, pbest i is to find the optimal initial cluster center position, gbest represents the global best cluster center, c1 and c2 are acceleration coefficients, and r1 and r2 are random numbers;

[0032] Repeat the above steps to calculate fitness and update particle positions until the maximum number of iterations is met; select the cluster center corresponding to the particle with the smallest fitness as the initial cluster center, and use multiple iterations of the K-means algorithm to assign the POI dataset to the nearest cluster center.

[0033] Furthermore, the method further comprises:

[0034] When the number of clusters grows close to the actual number of categories, the weighted variance ratio standard index and silhouette coefficient of each cluster obtained by clustering are calculated to analyze the internal and external relationships of the cluster. The optimal clustering result is selected according to the analysis results, and regional clustering is performed based on the best clustering result to achieve personalized interest area extraction.

[0035] One or more embodiments of this specification provide a personalized interest region extraction system based on interest units, including:

[0036] Interest unit division module: used for extracting urban road data in the study area, and dividing the study area into different interest units based on the urban road data;

[0037] Feature fusion module: Based on the check-in data of the interest unit, dynamically assign corresponding weight parameters to the convolution layer, extract the time interval feature and spatial distance feature of the interest unit, and fuse the above features to form a spatiotemporal fusion feature;

[0038] Interest area extraction module: taking the spatiotemporal fusion features as input, clustering the interest units through an improved clustering algorithm, dividing different interest areas, and realizing the extraction of personalized interest areas.

[0039] One or more embodiments of the present specification provide an electronic device, including:

[0040] A processor; and a memory arranged to store computer executable instructions; when the computer executable instructions are executed, the processor implements the steps of the above-mentioned method for extracting personalized interest areas based on interest units.

[0041] One or more embodiments of the present specification provide a storage medium for storing computer-executable instructions, which, when executed, implement the steps of the method for extracting personalized interest regions based on interest units as described in any one of claims 1 to 7.

[0042] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the prior art of this specification, the drawings required for use in the embodiments or the prior art description will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0044] Figure 1 A flowchart of a personalized interest area extraction method based on interest units provided for one or more embodiments of this specification;

[0045] Figure 2 A specific implementation flow chart of a method for extracting a personalized interest region based on an interest unit provided in one or more embodiments of this specification;

[0046] Figure 3 A schematic diagram of the composition of a personalized interest region extraction system based on interest units provided in one or more embodiments of this specification;

[0047] Figure 4 A schematic diagram of the structure of an electronic device provided for one or more embodiments of this specification. DETAILED DESCRIPTION

[0048] In order to enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the following will be combined with the drawings in one or more embodiments of this specification to clearly and completely describe the technical solutions in one or more embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this document.

[0049] Method Embodiment

[0050] According to an embodiment of the present invention, a method for extracting a personalized interest region based on an interest unit is provided. Figure 1 A flowchart of a personalized interest area extraction method based on interest units provided in one or more embodiments of this specification, Figure 2 A specific implementation flow chart of a personalized interest area extraction method based on interest units provided in one or more embodiments of this specification, such as Figure 1 and Figure 2 As shown, the personalized interest region extraction method based on interest units according to an embodiment of the present invention specifically includes:

[0051] S1. Extract urban road data in the study area, and divide the study area into different units of interest based on the urban road data.

[0052] Using the interest unit divided by the urban road data as the minimum unit of the user's interest area ROI can better capture the geographical relationship between the interest areas and play an important role in accurately extracting the ROI. The urban road data includes the highway, main road and urban main road data in OpenStreetMap.

[0053] The urban road data in the study area is extracted, and the study area is divided into different interest units based on the urban road data. The specific method is as follows:

[0054] For the study area, the data of highways, main roads and urban main roads in OpenStreetMap are selected as urban road data, and the selected urban road data are preprocessed to form new road data; specifically, the preprocessing of urban road data includes:

[0055] Remove duplicate, suspended and independent roads, and extend and complete disconnected roads; generate buffer zones of different ranges for roads of different levels, such as 40m, 20m and 10m buffer zones for first-level, second-level and third-level roads respectively; based on urban road data and administrative district data, remove noise and redundant information from urban road data, extract the overlapping parts of road and administrative district data through overlay analysis, and clean the overlapping areas of administrative district data as new road data.

[0056] When the new road data has topological errors and thus produces an irregular grid, the topological errors are modified, and the topological errors include the existence of hanging points, intersection points, etc. After the modification, the new road data forms a closed unit, and finally forms an interest unit based on the urban road data.

[0057] S2. Based on the check-in data of the interest unit, dynamically assign corresponding weight parameters to the convolution layer, extract the time interval features and spatial distance features of the interest unit, and fuse the above features to form a spatiotemporal fusion feature.

[0058] In urban areas, each unit of interest does not exist independently, but is interconnected with and affected by surrounding units. Therefore, ROI extraction is not only affected by the spatiotemporal characteristics of a single unit of interest, but also by surrounding units. Set the spatial graph G = {V, E, A}, where nodes V = {v0, v1, v2...v n-1} is the interest unit, each node v i Contains a d-dimensional feature {x1,x2,…,x d}, indicating the category information of all POIs (points of interest) in the interest unit; edge set E = {e0, e1, e3, …em-1}, each e i represents the edge connecting two nodes. If two interest units have a common edge, the corresponding two nodes are connected by an edge. The adjacency matrix A represents the spatial relationship between nodes. ij Indicates the unit of interest v i and v j Haversine distance between them.

[0059] In the encoding framework, based on the check-in data of interest units, the check-in frequency of different interest units is analyzed, and the activity of each interest unit is calculated by the check-in frequency. Based on the activity calculated by the check-in frequency of interest units, different convolution weights are dynamically assigned to the convolution layer, which are then applied to multi-layer graph convolution and dilated convolution networks to capture the complex spatiotemporal relationship between interest units.

[0060] After normalizing the adjacency matrix, the graph convolution operation is performed as follows:

[0061] Normalize the adjacency matrix:

[0062]

[0063] Perform graph convolution operation:

[0064]

[0065] Among them, X j is the feature vector of node j, and W is the category feature of the spatial graph node;

[0066] Introducing the convolution weights dynamically allocated according to the activity of each interest unit in each graph convolution layer;

[0067] By stacking multiple layers of graph convolutional networks, the network depth is increased to extract the spatial distance features of interest units;

[0068] The dilated convolutional network is introduced to expand the convolution receptive field by introducing a dilation factor in the time dimension. On this basis, the dynamic convolution mechanism is further introduced to dynamically adjust the convolution kernel weights to adapt to the feature changes at different time scales and extract the time interval features of the units of interest.

[0069] In the same dimensional space, a mapping relationship between spatial distance features and time interval features is established, and the spatial interval features and time interval features are fused in a homomorphic manner to obtain the spatiotemporal fusion feature f m .

[0070] In the decoding framework, first, f mInput to the decoder consisting of 4 convolutional layers; secondly, for each convolutional layer, use the transpose w of the convolution kernel corresponding to the layer to perform convolution operation, sum the convolution results; finally add the corresponding bias term. That is, calculate The result after fusion feature reconstruction is obtained to obtain spatiotemporal fusion features.

[0071] In the loss function, in order to prevent the model from overfitting and improve its generalization ability, L2 regularization is used in the form of Where λ is the regularization coefficient, and cross-validation is introduced to validate its λ value, ||W|| F is the Frobenius norm of the weight matrix.

[0072] S3. Taking the spatiotemporal fusion features as input, clustering the interest units through an improved clustering algorithm, dividing different interest areas, and realizing the extraction of personalized interest areas.

[0073] After extracting the spatiotemporal fusion features, a particle swarm algorithm is designed to optimize the cluster center. Unsupervised classification is used to divide similar interest units into the same area to achieve personalized user interest area extraction. Specifically:

[0074] The specific method of designing particle swarm algorithm to optimize cluster centers is as follows:

[0075] First, k particles are randomly generated, each particle contains the location of a possible initial cluster center;

[0076] For each particle, the fitness is calculated using the objective function of K-means clustering, where the objective function is to maximize the sum of the distance differences between the center cluster and the center of the cluster to which it belongs;

[0077] Calculate the particle cluster center position and moving direction to update the particle position. The specific calculation method is as follows:

[0078] x i (t+1)=x i (t)+v i (t+1);

[0079] v i (t+1)=w·v i (t)+c1·r1·(pbest i -x i (t))+c2·r2·(gbest-x i (t));

[0080] where x i (t) and v i (t) represent the particle cluster center position and moving direction, pbest iis to find the optimal initial cluster center position, gbest represents the global best cluster center, c1 and c2 are acceleration coefficients, and r1 and r2 are random numbers;

[0081] Repeat the above steps to calculate fitness and update particle positions until the maximum number of iterations is met; select the cluster center corresponding to the particle with the smallest fitness as the initial cluster center, and use multiple iterations of the K-means algorithm to assign the ROI data set to the nearest cluster center.

[0082] As the number of clusters increases, the degree of aggregation of each cluster increases significantly, so the variance ratio standard index also increases. However, when the number of clusters grows close to the number of true categories, further increasing the number of clusters will not significantly improve the degree of aggregation, causing the growth rate of the variance ratio standard index to gradually slow down and eventually level off. Considering the importance differences of clustering result categories, the result categories are introduced as weights into the formula of the variance ratio standard to improve the accuracy and stability of the clustering results. Specifically:

[0083] Calculate the weighted variance ratio standard index (WCH) as follows:

[0084]

[0085] Among them, n is the number of ROI samples, k is the number of clusters, and w B ,w k Represent the inter-cluster and intra-cluster weights respectively;

[0086] in:

[0087]

[0088] B k is the weighted covariance matrix between ROI clusters, m i and w i are the center point and weight factor of the i-th cluster respectively; W is the weighted covariance matrix of the ROI within the cluster, tr is the trace of the matrix, m i and w i are the center point and weight factor of the i-th cluster respectively; therefore, in WCH k The figure shows a trend of slowing down the rise, and the k value corresponding to the inflection point is the optimal number of clusters;

[0089] Calculate the silhouette coefficient to evaluate the overall distribution of clustering quality. The specific calculation method is as follows:

[0090]

[0091] SC(i) represents the silhouette coefficient of the i-th cluster, a(i) and b(i) represent the average distance between the i-th cluster and other samples in the same cluster and the average distance between the i-th cluster and the samples in the nearest other clusters, respectively. When SC(i) is close to 1, it means that the ROI sample clustering is reasonable; when SC(i) is close to -1, it means that the ROI set classification is unreasonable and should be classified as other clusters; when SC(i) is close to 0, it means that the ROI set is located at the boundary of two clusters. Therefore, the k value corresponding to SC(i) close to 1 is taken as the best clustering quality.

[0092] Finally, the optimal clustering result is selected according to the analysis results of the optimal number of clusters and the optimal clustering quality, and regional clustering is performed based on the optimal clustering result to achieve personalized interest area extraction.

[0093] System Example

[0094] According to an embodiment of the present invention, a personalized interest region extraction system based on interest units is provided. Figure 3 A schematic diagram of a composition of a personalized interest region extraction system based on interest units provided in one or more embodiments of this specification, such as Figure 3 As shown, the personalized interest region extraction system based on interest units according to an embodiment of the present invention specifically includes:

[0095] The unit of interest division module 30 is used to extract urban road data in the study area, and divide the study area into different units of interest based on the urban road data;

[0096] Feature fusion module 32: used to dynamically assign corresponding weight parameters to the convolution layer based on the check-in data of the interest unit, extract the time interval feature and spatial distance feature of the interest unit, and fuse the above features to form a spatiotemporal fusion feature;

[0097] The interest region extraction module 34 takes the spatiotemporal fusion features as input, clusters the interest units through an improved clustering algorithm, divides different interest regions, and realizes the extraction of personalized interest regions.

[0098] The embodiment of the present invention is a system embodiment corresponding to the above-mentioned method embodiment. The specific operations of each module can be understood by referring to the description of the method embodiment, which will not be repeated here.

[0099] Device Example 1

[0100] An embodiment of the present invention provides an electronic device, such as Figure 4 As shown, it includes: a memory 40, a processor 42, and a computer program stored in the memory 40 and executable on the processor 42. When the computer program is executed by the processor 42, the following method steps are implemented:

[0101] S1. extracting urban road data within the study area, and dividing the study area into different interest units based on the urban road data;

[0102] S2. Based on the check-in data of the interest unit, dynamically assign corresponding weight parameters to the convolution layer, extract the time interval feature and spatial distance feature of the interest unit, and fuse the above features to form a spatiotemporal fusion feature;

[0103] S3. Taking the spatiotemporal fusion features as input, clustering the interest units through an improved clustering algorithm, dividing different interest areas, and realizing the extraction of personalized interest areas.

[0104] Device Example 2

[0105] An embodiment of the present invention provides a computer-readable storage medium, on which a program for implementing information transmission is stored. When the program is executed by the processor 42, the following method steps are implemented:

[0106] S1. extracting urban road data within the study area, and dividing the study area into different interest units based on the urban road data;

[0107] S2. Based on the check-in data of the interest unit, dynamically assign corresponding weight parameters to the convolution layer, extract the time interval feature and spatial distance feature of the interest unit, and fuse the above features to form a spatiotemporal fusion feature;

[0108] S3. Taking the spatiotemporal fusion features as input, clustering the interest units through an improved clustering algorithm, dividing different interest areas, and realizing the extraction of personalized interest areas.

[0109] The computer-readable storage medium in this embodiment includes, but is not limited to, ROM, RAM, magnetic disk or optical disk, etc.

[0110] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A personalized interest region extraction method based on interest units, It is characterized in that include: Extracting urban road data within a study area, and dividing the study area into different units of interest based on the urban road data; Based on the check-in data of the interest unit, dynamically assign corresponding weight parameters to the convolution layer, extract the time interval feature and spatial distance feature of the interest unit, and fuse the above features to form a spatiotemporal fusion feature; The spatiotemporal fusion features are used as input, and the interest units are clustered through an improved clustering algorithm to divide different interest areas, thereby realizing the extraction of personalized interest areas.

2. The method according to claim 1, It is characterized in that The urban road data includes data on expressways, main roads and urban main roads in OpenStreetMap; Extracting urban road data within the study area, and dividing the study area into different interest units based on the urban road data specifically includes: For the study area, the highway, main road and urban trunk road data in OpenStreetMap were selected as urban road data; Preprocess the selected urban road data to form new road data; When the new road data has topological errors and thus generates irregular grids, the topological errors are modified so that the new road data forms closed units, and finally forms interest units based on the urban road data.

3. The method according to claim 2, It is characterized in that Preprocessing the selected urban road data to form new road data specifically includes: Remove duplicate, hanging and independent roads, and extend and complete disconnected roads; Generate buffer zones of different ranges for roads of different levels; Based on urban road data and administrative district data, the noise and redundant information of urban road data are removed, the overlapping parts of road and administrative district data are integrated through overlay analysis, and the overlapping areas of administrative district data are cleaned.

4. The method according to claim 1, It is characterized in that Based on the check-in data of the interest unit, the corresponding weight parameters are dynamically assigned to the convolution layer to extract the time interval features and spatial distance features of the interest unit, specifically including: Set the spatial graph G = {V, E, A}, where node V = {v 0 ,v 1 ,v 2 ……v n-1 } is the interest unit, and the edge set E = {e 0 ,e 1 ,e 3 ,…e m-1 }, each e i represents the edge connecting two nodes, and the adjacency matrix A represents the spatial relationship between nodes. ij Indicates the unit of interest v i and v j Haversine distance between them; Based on the check-in data of interest units, the check-in frequency of different interest units is analyzed, and the activity of each interest unit is calculated according to the check-in frequency; according to the activity of each interest unit, different convolution weights are dynamically assigned to the convolution layer, and then applied to multi-layer graph convolution and dilated convolution networks to capture the complex spatiotemporal relationship between interest units; After normalizing the adjacency matrix, a graph convolution operation is performed, and the convolution weights are introduced into each graph convolution layer. Multiple layers of graph convolution are stacked to increase the network depth and extract the spatial distance features of the units of interest. The dilated convolutional network is introduced to expand the convolution receptive field by introducing a dilation factor in the time dimension. On this basis, a dynamic convolution mechanism is further introduced to dynamically adjust the convolution kernel weights to adapt to the feature changes at different time scales and extract the time interval features of the units of interest.

5. The method according to claim 4, It is characterized in that The specific method of fusing spatial distance features and time interval features to form spatiotemporal fusion features is as follows: In the same dimensional space, a mapping relationship between spatial distance features and time interval features is established, and the spatial interval features and time interval features are fused in a homomorphic manner to obtain the spatiotemporal fusion feature f m .

6. The method according to claim 1, It is characterized in that The spatiotemporal fusion features are used as input, and the interest units are clustered by an improved clustering algorithm to divide different interest areas, so as to realize the extraction of personalized interest areas. The specific method is as follows: A particle swarm algorithm is designed to optimize cluster centers and unsupervised classification is used to divide similar units of interest into the same area. The specific method of designing particle swarm algorithm to optimize cluster centers is as follows: First, k particles are randomly generated, each particle contains the location of a possible initial cluster center; For each particle, the fitness is calculated using the objective function of K-means clustering, where the objective function is to maximize the sum of the distance differences between the center cluster and the center of the cluster to which it belongs; Calculate the particle cluster center position and moving direction to update the particle position. The specific calculation method is as follows: x i (t+1)=x i (t)+v i (t+1); v i (t+1)=w·v i (t)+c 1 ·r 1 ·(pbest i -x i (t))+c 2 ·r 2 ·(gbest-x i (t)); where x i (t) and v i (t) represent the particle cluster center position and moving direction, pbest i is to find the optimal initial cluster center position, gbest represents the global best cluster center, c 1 、c 2 is the acceleration factor, r 1 、r 2 is a random number; Repeat the above steps to calculate fitness and update particle positions until the maximum number of iterations is met; The cluster center corresponding to the particle with the smallest fitness is selected as the initial cluster center, and the POI dataset is assigned to the nearest cluster center using multiple iterations of the K-means algorithm.

7. The method according to claim 6, It is characterized in that The method further comprises: When the number of clusters grows close to the actual number of categories, the weighted variance ratio standard index and silhouette coefficient of each cluster obtained by clustering are calculated to analyze the internal and external relationships of the cluster. The optimal clustering result is selected according to the analysis results, and regional clustering is performed based on the best clustering result to achieve personalized interest area extraction.

8. A personalized interest region extraction system based on interest units, It is characterized in that include: Interest unit division module: used for extracting urban road data in the study area, and dividing the study area into different interest units based on the urban road data; Feature fusion module: Based on the check-in data of the interest unit, dynamically assign corresponding weight parameters to the convolution layer, extract the time interval feature and spatial distance feature of the interest unit, and fuse the above features to form a spatiotemporal fusion feature; Interest area extraction module: taking the spatiotemporal fusion features as input, clustering the interest units through an improved clustering algorithm, dividing different interest areas, and realizing the extraction of personalized interest areas.

9. An electronic device, It is characterized in that include: processor; and a memory arranged to store computer executable instructions; when the computer executable instructions are executed, the processor implements the steps of the personalized interest area extraction method based on interest units as described in any one of claims 1 to 7.

10. A storage medium, It is characterized in that Used to store computer executable instructions, which, when executed, implement the steps of the personalized interest area extraction method based on interest units as described in any one of claims 1 to 7.

Citation Information

Patent Citations

  • Checkin data based spatial-temporal trajectory similarity measurement method and system

    CN105243148A

  • City vitality measurement and characterization method and system based on multi-source big data

    CN112561401A

  • Interest point recommendation method based on graph neural network and collaborative relation mining

    CN116738051A

  • KR20190124524A