Similar person identification method, device, equipment and storage medium

Abstract

Embodiments of the present application disclose a similar crowd identification method and device, equipment and storage medium, wherein the method comprises: acquiring a plurality of signaling data and signaling attributes of each signaling data; the signaling attributes comprise location information and initial users corresponding to the signaling data; performing multi-level gridding processing on the plurality of signaling data based on the location information of each signaling data, obtaining correlation degree information between the initial users, and determining a first similar crowd set similar to a seed user set; based on a trained similar crowd identification model corresponding to the seed user set, screening a plurality of initial users corresponding to the plurality of signaling data to obtain a second similar crowd set similar to the seed user set; and taking the intersection of the first similar crowd set and the second similar crowd set as a target similar crowd set.

Description

Technical Field

[0001] This application relates to, but is not limited to, the technical field, and in particular to a method, apparatus, device, and storage medium for identifying similar groups of people. Background Technology

[0002] Currently, communication and internet technologies have developed rapidly, leading to an explosive growth of big data in the information age. Data mining and value discovery within this big data are in full swing. Among these efforts, the subcategory of identifying similar user groups within massive amounts of data has seen significant and valuable technological research. This identification method has a wide range of essential applications in areas such as user discovery, targeted advertising, targeted information dissemination by public management departments, and targeted anti-fraud campaigns. Summary of the Invention

[0003] In view of this, embodiments of this application provide at least one method, apparatus, device, and storage medium for identifying similar groups of people.

[0004] The technical solution of this application embodiment is implemented as follows:

[0005] On one hand, embodiments of this application provide a method for identifying similar groups of people, the method comprising:

[0006] Obtain multiple signaling data and signaling attributes for each signaling data; the signaling attributes include location information and the initial user corresponding to the signaling data;

[0007] Based on the location information of each signaling data, the multiple signaling data are processed by multi-level rasterization to obtain the correlation information between each initial user, and to determine the first similar group set similar to the seed user set.

[0008] Based on the trained similar user identification model corresponding to the seed user set, multiple initial users corresponding to the multiple signaling data are screened to obtain a second similar user set similar to the seed user set;

[0009] The intersection of the first set of similar people and the second set of similar people is taken as the target set of similar people.

[0010] In some embodiments, the step of performing multi-level rasterization processing on the plurality of signaling data based on the location information of each of the signaling data to obtain the correlation information between the initial users and to determine a first similar group set similar to the seed user set includes:

[0011] Based on the location information of each signaling data, the plurality of signaling data are subjected to multiple levels of rasterization processing to obtain the raster division result corresponding to each level; the raster division result includes the raster area corresponding to each signaling data at the level;

[0012] For each level, based on the initial user and grid area corresponding to each signaling data, an initial user relationship graph corresponding to that level is determined; the initial user relationship graph represents whether there is a relationship between each initial user;

[0013] A comprehensive association graph is generated based on the level weight corresponding to each level and the initial user association graph corresponding to each level; the comprehensive association graph includes the association degree information between each initial user.

[0014] Based on the correlation information between the initial users in the comprehensive correlation graph, abnormal users are removed from the multiple initial users to obtain a first similar group set that is similar to the seed user set.

[0015] In some embodiments, the rasterization result includes the raster region corresponding to each signaling data at the level; the process of performing rasterization processing on the plurality of signaling data at multiple levels based on the location information of each signaling data to obtain the rasterization result corresponding to each level includes:

[0016] For each level of rasterization processing, obtain the set of raster regions corresponding to that level;

[0017] Based on the location information of each signaling data and each grid region in the grid region set, the grid region corresponding to each signaling data at the level is determined.

[0018] In some embodiments, determining the initial user association graph corresponding to the level based on the initial user and grid region corresponding to each signaling data includes:

[0019] Based on the association between the initial user and the grid region in each of the signaling data, a bipartite graph is constructed; the bipartite graph represents whether there is an association between each initial user and each grid region;

[0020] The bipartite graph is converted into the initial user association graph corresponding to the level.

[0021] In some embodiments, constructing a bipartite graph based on the association between the initial user and the grid region in each of the signaling data includes:

[0022] Each of the grid regions is taken as a grid vertex in the bipartite graph, and each of the initial users is taken as the first user vertex in the bipartite graph;

[0023] For each of the signaling data, a first connection relationship is established between the grid vertex corresponding to the signaling data and the first user vertex to obtain the bipartite graph; the first connection relationship indicates that there is an association between the grid region corresponding to the grid vertex and the initial user corresponding to the first user vertex.

[0024] In some embodiments, converting the bipartite graph into an initial user association graph corresponding to the level includes:

[0025] For each grid vertex in the bipartite graph, an initial user set corresponding to the grid vertex is obtained in the bipartite graph; the initial user set includes the initial users corresponding to all first user vertices that have a first connection relationship with the grid vertex;

[0026] Each of the initial users is designated as the second user vertex in the initial user association graph;

[0027] For each grid vertex corresponding to the initial user set, a second connection relationship is established between the second user vertices corresponding to every two initial users in the initial user set, resulting in an initial user association graph corresponding to the level; the second connection relationship indicates that the initial users corresponding to the two second user vertices have an association relationship.

[0028] In some embodiments, the initial user association graph includes second user vertices and at least one second connection relationship, wherein the second connection relationship indicates that the initial users corresponding to two second user vertices are associated; generating a comprehensive association graph based on the level weights corresponding to each level and the initial user association graphs corresponding to each level includes:

[0029] Based on the union of the second user vertices in the initial user association graph corresponding to each level, the third user vertex in the comprehensive association graph is constructed.

[0030] Based on the union of at least one second connection relationship in the initial user association graph corresponding to each level, at least one third connection relationship is obtained in the comprehensive association graph; the third connection relationship indicates that the initial users corresponding to the two third user vertices have an association relationship.

[0031] Based on the level weight corresponding to each level and at least one second connection relationship in the initial user association graph corresponding to each level, the weight sum corresponding to each third connection relationship in the comprehensive association graph is determined; the weight sum represents the association degree information between the two initial users.

[0032] In some embodiments, the abnormal users include other users and seed users in the seed user set; the step of removing abnormal users from the plurality of initial users based on the correlation information between the initial users in the comprehensive correlation graph to obtain a first similarity group set similar to the seed user set includes:

[0033] Based on the correlation information between the initial users in the comprehensive correlation graph, the initial users who have no connection relationship with each of the seed users are selected as the other users among the plurality of initial users;

[0034] The other users and the seed users are removed from the plurality of initial users to obtain a first similarity group set that is similar to the seed user set.

[0035] In some embodiments, the integrated association graph includes a third user vertex, at least one third connection relationship, and a weight sum corresponding to each third connection relationship, wherein the third connection relationship indicates that the initial users corresponding to two corresponding third user vertices have an association relationship; the method further includes:

[0036] The comprehensive association graph retains the third user vertex corresponding to the seed user in the seed user set and the third user vertex corresponding to the target similar user in the target similar group set;

[0037] For each target similar user, the weights corresponding to the third connection relationship between the third user vertex corresponding to the target similar user and the third user vertex corresponding to the seed user are summed to obtain the similarity between the target similar user and the set of seed users.

[0038] Based on the similarity of each target similar user, a target similar group set with ranking information is generated.

[0039] In some embodiments, the step of filtering multiple initial users corresponding to the multiple signaling data based on the trained similar user group identification model corresponding to the seed user set to obtain a second similar user group set similar to the seed user set includes:

[0040] Obtain the profile data corresponding to each of the initial users corresponding to the plurality of signaling data, as a profile dataset;

[0041] Based on the aforementioned profile dataset, a model is trained to obtain a similar population recognition model corresponding to the seed user set.

[0042] Remove the signaling data from the multiple signaling data sets where the initial user is a seed user, to obtain the remaining user set;

[0043] Based on the trained similar user identification model corresponding to the seed user set, the remaining user set is filtered to obtain a second similar user set that is similar to the seed user set.

[0044] On the other hand, embodiments of this application provide a similar crowd identification device, the device comprising:

[0045] An acquisition module is used to acquire multiple signaling data and signaling attributes of each signaling data; the signaling attributes include location information and the initial user corresponding to the signaling data;

[0046] The first similar group determination module is used to perform multi-level rasterization processing on the multiple signaling data based on the location information of each signaling data to obtain the correlation information between each initial user, and determine the first similar group set similar to the seed user set;

[0047] The second similar user group determination module is used to filter multiple initial users corresponding to the multiple signaling data based on the trained similar user group identification model corresponding to the seed user set, and obtain a second similar user group set similar to the seed user set.

[0048] The target similar population determination module is used to take the intersection of the first similar population set and the second similar population set as the target similar population set.

[0049] In another aspect, embodiments of this application provide a computer device, including a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the program to implement some or all of the steps in the above-described method.

[0050] In another aspect, embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements some or all of the steps in the above-described method.

[0051] In this embodiment, a first set of similar groups is obtained by simultaneously using multi-level rasterization processing, and a second set of similar groups is obtained by using a similar group recognition model. Then, the first and second sets of similar groups are fused by finding their intersection. In this way, compared with the traditional technology that only uses machine learning methods to identify similar groups, this embodiment can fully explore the spatiotemporal relationship of signaling data by using multi-rasterization processing. It can also utilize the characteristics of signaling data (that is, people with similar attributes within a spatiotemporal range have more similar attributes) to obtain the first set of similar groups. At the same time, the fusion method in this embodiment improves the accuracy of similar group identification and the density of the identified similar groups.

[0052] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and are not intended to limit the technical solutions of this application. Attached Figure Description

[0053] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with this application and, together with the specification, serve to explain the technical solutions of this application.

[0054] Figure 1 A schematic diagram illustrating the implementation process of a similar group identification method provided in this application embodiment;

[0055] Figure 2A A schematic diagram illustrating the implementation process of a similar group identification method provided in this application embodiment;

[0056] Figure 2B This is a schematic diagram illustrating the construction of a bipartite graph provided in an embodiment of this application;

[0057] Figure 2C This application provides a schematic diagram illustrating the construction of an initial user association graph.

[0058] Figure 2D This application provides a schematic diagram of an initial user association graph corresponding to another level.

[0059] Figure 2E A schematic diagram of a comprehensive correlation diagram provided for an embodiment of this application;

[0060] Figure 3 A schematic diagram illustrating the implementation process of a similar group identification method provided in this application embodiment;

[0061] Figure 4 A schematic diagram illustrating the implementation process of a similar group identification method provided in this application embodiment;

[0062] Figure 5 A schematic diagram illustrating the implementation process of a similar group identification method provided in this application embodiment;

[0063] Figure 6 This is a schematic diagram illustrating the calculation process of a similar population dataset A provided in an embodiment of this application;

[0064] Figure 7 A schematic diagram of multi-level raster conversion provided in an embodiment of this application;

[0065] Figure 8 A schematic diagram of bipartite graph connections provided in an embodiment of this application;

[0066] Figure 9 This application provides a schematic diagram illustrating the relationship between merged mobile phone users.

[0067] Figure 10 A schematic diagram illustrating the calculation process of a similar population dataset B provided in an embodiment of this application;

[0068] Figure 11 A schematic diagram illustrating the implementation process of similarity ranking provided in an embodiment of this application;

[0069] Figure 12 This is a schematic diagram of the composition structure of a similar crowd identification device provided in an embodiment of this application;

[0070] Figure 13 This is a schematic diagram of the hardware entity of a computer device provided in an embodiment of this application. Detailed Implementation

[0071] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application are further described in detail below with reference to the accompanying drawings and embodiments. The described embodiments should not be regarded as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0072] In the following description, references to "some embodiments" refer to a subset of all possible embodiments. It is understood that "some embodiments" may be the same or different subsets of all possible embodiments and may be combined with each other without conflict. The terms "first / second / third" are used merely to distinguish similar objects and do not represent a specific ordering of objects. It is understood that "first / second / third" may be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0073] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. The terminology used herein is for descriptive purposes only and is not intended to limit the scope of this application.

[0074] This application provides a method for identifying similar groups of people, which can be executed by the processor of a computer device. The computer device refers to a device with data processing capabilities, such as a server, laptop, tablet, desktop computer, smart TV, set-top box, or mobile device (e.g., mobile phone, portable video player, personal digital assistant, dedicated messaging device, portable gaming device).

[0075] Figure 1 This is a schematic diagram illustrating the implementation process of a similar group identification method provided in an embodiment of this application, as shown below. Figure 1As shown, the method includes the following steps S101 to S104:

[0076] Step S101: Obtain multiple signaling data and the signaling attributes of each signaling data; the signaling attributes include location information and the initial user corresponding to the signaling data.

[0077] In some embodiments, for the acquired original signaling data set, which may include all acquired signaling data, in order to improve the identification efficiency and accuracy of similar groups, the multiple signaling data can be acquired from the original signaling data set first, and then the target similar group set similar to the seed user set can be determined from the multiple signaling data.

[0078] In some embodiments, the signaling attribute information of the signaling data also includes the timestamp of the signaling generation. Then, by setting a time range, signaling data with timestamps within the time range can be selected from the original signaling data set to obtain the multiple signaling data.

[0079] In some embodiments, signaling data whose location information is located within the range of the original signaling data set can be selected by setting a range, thereby obtaining the plurality of signaling data.

[0080] In some embodiments, the time range and the area range can be set simultaneously, and then signaling data whose location information is within the area range and whose timestamp is within the time range can be selected from the original signaling data set to obtain the multiple signaling data.

[0081] In some embodiments, the location information may include at least one of the following: a unique identifier for the base station, the administrative district / county code to which the base station belongs, the latitude of the base station, and the longitude of the base station.

[0082] In some embodiments, the initial user corresponding to the signaling data may be represented in the form of an International Mobile Subscriber Identity (IMSI) and / or a user's mobile phone number.

[0083] Step S102: Perform multi-level rasterization processing on the multiple signaling data based on the location information of each signaling data to obtain the correlation information between each initial user, and determine the first similar group set similar to the seed user set.

[0084] In some embodiments, the seed user set is a set of users for which a corresponding similar group needs to be found. The seed user set may include multiple seed users. In this application embodiment, the purpose is actually to find a similar group to the group formed by the multiple seed users among the multiple initial users corresponding to multiple signaling data. The multiple initial users corresponding to the multiple signaling data may include the seed users.

[0085] In some embodiments, at least two levels of rasterization processing can be set up, thereby obtaining at least two raster division results. Each level of rasterization processing can determine, based on location information and the raster region divided at the current level, whether the initial users in the current level are located within the same raster region, thus generating a raster division result for each level. Then, combining the raster division results at each level, correlation information between the initial users is generated. This correlation information is used at least to characterize whether there is a correlation between the corresponding two initial users.

[0086] In some embodiments, since the seed user is included among the multiple initial users corresponding to the multiple signaling data, and the correlation information between each initial user is at least used to characterize whether there is a correlation between the two initial users, a first similar user similar to various sub-users can be found among the multiple initial users to form the first similar group set.

[0087] Step S103: Based on the trained similar user group identification model corresponding to the seed user set, filter the multiple initial users corresponding to the multiple signaling data to obtain a second similar user group set similar to the seed user set.

[0088] In some embodiments, a look-alike technique can be used to determine a second set of similar users from among the multiple initial users corresponding to the multiple signaling data sets, based on the seed user set. The look-alike technique refers to a technique that, based on a set of seed users, uses an algorithmic model to find groups of people similar to or potentially related to the seed users.

[0089] In other embodiments, seed user features corresponding to various sub-users in the seed user set can be extracted. Based on the seed user features corresponding to each seed user, at least one central feature corresponding to the seed user set can be generated. Then, based on this at least one central feature, a second similarity group set similar to the seed user set is obtained by filtering among the multiple initial users. The at least one central feature can be clustered using a clustering algorithm to obtain at least one seed user cluster. For each seed user cluster, a central feature can be obtained by fusing the seed user features corresponding to various sub-users within that seed user cluster.

[0090] Step S104: Take the intersection of the first set of similar people and the second set of similar people as the target set of similar people.

[0091] In some embodiments, the first similar user set, obtained after the multi-level rasterization processing described above, is a similar user group determined based on the location information and raster division of signaling data. The second similar user set, obtained based on the similar user identification model described above, is a similar user group determined based on the user characteristics of each user. Both sets of similar users are determined by referencing different feature dimensions. Determining the target similar user set by finding their intersection can improve the accuracy of similar user identification.

[0092] In this embodiment, a first set of similar groups is obtained by simultaneously using multi-level rasterization processing, and a second set of similar groups is obtained by using a similar group recognition model. Then, the first and second sets of similar groups are fused by finding their intersection. In this way, compared with the traditional technology that only uses machine learning methods to identify similar groups, this embodiment can fully explore the spatiotemporal relationship of signaling data by using multi-rasterization processing. It can also utilize the characteristics of signaling data (that is, people with similar attributes within a spatiotemporal range have more similar attributes) to obtain the first set of similar groups. At the same time, the fusion method in this embodiment improves the accuracy of similar group identification and the density of the identified similar groups.

[0093] Figure 2A This is an optional flowchart illustrating a similar crowd identification method provided in an embodiment of this application. This method can be executed by the processor of a computer device. Based on... Figure 1 , Figure 1 Step S102 can be updated to steps S201 to S204, combining Figure 2A The steps shown are explained.

[0094] Step S201: Based on the location information of each signaling data, perform multiple levels of rasterization processing on the multiple signaling data to obtain the raster division result corresponding to each level; the raster division result includes the raster area corresponding to each signaling data at the level.

[0095] In some embodiments, for any level of rasterization, the multiple signaling data can be divided according to the raster region set corresponding to the current level. That is, based on the location information of each signaling data, the raster region in which each signaling data falls within the raster region set corresponding to the current level is determined, thereby obtaining the raster division result corresponding to the current level. Taking the number of levels as N, where N is an integer greater than or equal to 2, rasterization of N levels can yield raster division results for N levels.

[0096] In some embodiments, steps S2011 to S2012 can be used to perform multiple levels of rasterization processing on the plurality of signaling data based on the location information of each signaling data, so as to obtain the raster division result corresponding to each level.

[0097] Step S2011: For each level of rasterization processing, obtain the set of raster regions corresponding to that level.

[0098] In some embodiments, the criteria for dividing the set of raster regions corresponding to different levels of rasterization processing are different. The difference may be a difference in the size of the raster regions, a difference in the position of the raster regions, or a difference in the shape of the raster regions. This application does not limit this.

[0099] Step S2012: Based on the location information of each signaling data and each grid region in the grid region set, determine the grid region corresponding to each signaling data at the level.

[0100] In some embodiments, for each signaling data, the location information of the signaling data and the area information of each grid area in the grid area set can be used to determine which grid area the signaling data falls into, and the grid area into which it falls can be used as the grid area corresponding to the signaling data at the level.

[0101] Step S202: For each level, based on the initial user and grid area corresponding to each signaling data, determine the initial user association graph corresponding to the level; the initial user association graph represents whether there is an association relationship between each initial user.

[0102] In the initial user association graph corresponding to a certain level, two initial users with an association relationship must fall into the same grid area corresponding to that level.

[0103] In some embodiments, for any given grid area, if at least two signaling data points fall into that grid area, and these at least two signaling data points correspond to at least two initial users, it indicates that the at least two initial users are located in the same area within a defined spatiotemporal range, and thus it is necessary to establish an association between the at least two initial users. Similarly, based on the signaling data falling into each grid area and the initial user corresponding to each signaling data point, it is possible to determine whether an association exists between the initial users.

[0104] In some embodiments, steps S2021 to S2022 can be used to determine the initial user association graph corresponding to the level based on the initial user and grid area corresponding to each signaling data.

[0105] Step S2021: Based on the association between the initial user and the grid region in each of the signaling data, construct a bipartite graph; the bipartite graph represents whether there is an association between each initial user and each of the grid regions.

[0106] The step of constructing a bipartite graph based on the association between the initial user and the grid region in each of the signaling data includes: taking each grid region as a grid vertex in the bipartite graph and taking each initial user as a first user vertex in the bipartite graph; for each of the signaling data, establishing a first connection relationship between the grid vertex corresponding to the signaling data and the first user vertex to obtain the bipartite graph; the first connection relationship indicates that there is an association between the grid region corresponding to the grid vertex and the initial user corresponding to the first user vertex.

[0107] For example, please refer to Figure 2B The diagram illustrates the construction of a bipartite graph. If initial user m is associated with grid regions a and c, and initial user n is associated with grid regions b and c, then grid regions a, b, and c can be considered as grid vertices in the bipartite graph, and initial users m and n as the first user vertices. Then, the first connections between initial user m and grid region a, initial user m and grid region c, initial user n and grid region b, and initial user n and grid region c are established respectively. This leads to the following... Figure 2B The bipartite graph shown.

[0108] Step S2022: Convert the bipartite graph into the initial user association graph corresponding to the level.

[0109] Specifically, for each grid vertex in the bipartite graph, an initial user set corresponding to the grid vertex is obtained in the bipartite graph; the initial user set includes initial users corresponding to all first user vertices that have a first connection relationship with the grid vertex; each initial user is used as a second user vertex in the initial user association graph; for each initial user set corresponding to a grid vertex, a second connection relationship is established between the second user vertices corresponding to every two initial users in the initial user set to obtain the initial user association graph corresponding to the level; the second connection relationship indicates that the initial users corresponding to the two second user vertices have an association relationship.

[0110] For example, please refer to Figure 2C It illustrates a schematic diagram of the construction of an initial user association graph. Based on Figure 2BThe illustrated bipartite graph, for grid vertex a, obtains the initial user set (m) corresponding to grid vertex a in the bipartite graph; for grid vertex b, obtains the initial user set (n) corresponding to grid vertex b in the bipartite graph; for grid vertex c, obtains the initial user set (m, n) corresponding to grid vertex b in the bipartite graph; initial user m and initial user n are used as second user vertices in the initial user association graph; for initial user sets (m, n), establish a second connection relationship between the second user vertices corresponding to initial user m and initial user n; for initial user sets (m) and (n), since there is only one initial user, no corresponding second connection relationship is established. Afterwards, the following can be obtained: Figure 2C The initial user association graph corresponding to this level is shown.

[0111] Step S203: Generate a comprehensive association graph based on the level weight corresponding to each level and the initial user association graph corresponding to each level; the comprehensive association graph includes the association degree information between each initial user.

[0112] In some embodiments, the initial user association graph includes second user vertices and at least one second connection relationship, wherein the second connection relationship indicates that the initial users corresponding to two second user vertices are associated. The above-mentioned generation of a comprehensive association graph based on the level weights corresponding to each level and the initial user association graph corresponding to each level can be achieved through steps S2031 to S2033.

[0113] Step S2031: Based on the union of the second user vertices in the initial user association graph corresponding to each level, construct the third user vertex in the comprehensive association graph.

[0114] For example, this section illustrates the generation method of a comprehensive association graph using an initial user association graph with two levels; the number of levels is not limited. Please refer to [link / reference]. Figure 2D This shows the initial user association graph corresponding to another level. Step S2031 above includes: based on... Figure 2C The second user vertex (m, n) in the initial user association graph shown, and Figure 2D The second user vertex (m, n, o, p, q) in the initial user association graph is shown. The union of these two vertices (m, n, o, p, q) is constructed as the third user vertex in the comprehensive association graph, as follows: Figure 2E As shown.

[0115] Step S2032: Based on the union of at least one second connection relationship in the initial user association graph corresponding to each level, at least one third connection relationship is obtained in the comprehensive association graph; the third connection relationship indicates that the initial users corresponding to the two third user vertices have an association relationship.

[0116] Based on the above example, based on Figure 2C The second connection relationship "mn" in the and based on Figure 2D The union of the second connection relations "qn", "pn", "mn", "mp", and "mo" is used to construct the corresponding connection relations among the third user vertices (m, n, o, p, q), resulting in the following: Figure 2E The comprehensive correlation diagram shown shows that, due to... Figure 2C The second connection in is Figure 2D The subset of the second connection relation in, therefore Figure 2E The comprehensive correlation diagram shown is as follows Figure 2D same.

[0117] Step S2033: Based on the level weight corresponding to each level and at least one second connection relationship in the initial user association graph corresponding to each level, determine the weight sum corresponding to each third connection relationship in the comprehensive association graph; the weight sum represents the association degree information between the two initial users.

[0118] Based on the above example, if Figure 2C The weight corresponding to the level is 2. Figure 2D If the weight corresponding to the level is 1, then in the comprehensive association graph shown in 2E, the weight sum of the third connection relationship "qn" is 1, the weight sum of the third connection relationship "pn" is 1, the weight sum of the third connection relationship "mn" is 3, the weight sum of the third connection relationship "mp" is 1, and the weight sum of the third connection relationship "mo" is 1.

[0119] Step S204: Based on the correlation information between the initial users in the comprehensive correlation graph, remove abnormal users from the multiple initial users to obtain a first similar group set that is similar to the seed user set.

[0120] In some embodiments, steps S2041 to S2042 can be used to remove abnormal users from the plurality of initial users based on the correlation information between the initial users in the comprehensive correlation graph, thereby obtaining a first similar group set that is similar to the seed user set.

[0121] Step S2041: Based on the correlation information between the initial users in the comprehensive correlation graph, the initial users who do not have a connection relationship with each of the seed users are selected as the other users.

[0122] Step S2042: Remove the other users and the seed users from the plurality of initial users to obtain a first similar group set that is similar to the seed user set.

[0123] Based on the example above, if the seed users include n and p, then Figure 2E In the comprehensive association diagram shown, o represents other users, and n and p represent seed users. After removing the other users and seed users, the first similar group set can be obtained, which includes (m, q).

[0124] In this embodiment, multiple levels of rasterization are used to obtain an initial user association map for each level. Compared with related technologies that determine user association based on a single base station dimension, this avoids situations where different users are actually close but fail to be effectively identified because they do not have signaling data under the same base station. Furthermore, using multi-layer rasterization, compared with a single base station, enables personnel identification within multiple ranges. Simultaneously, a comprehensive association map is generated based on the level weights corresponding to each level and the initial user association maps corresponding to each level. Thus, the longer two users stay in similar locations, the more associations they have in the multi-layer rasterization, and the more connections will be present in the corresponding multi-layer initial user association maps. After merging to obtain the comprehensive association map, the association information of these two users is also higher, improving the accuracy of identifying similar groups of people.

[0125] Figure 3 This is an optional flowchart illustrating a similar crowd identification method provided in an embodiment of this application. This method can be executed by the processor of a computer device. Based on... Figure 1 , Figure 1 Step S103 can be updated to steps S301 to S304, combining Figure 3 The steps shown are explained.

[0126] Step S301: Obtain the profile data corresponding to each of the initial users corresponding to the plurality of signaling data, as a profile dataset.

[0127] Step S302: Train the model based on the portrait dataset to obtain the trained similar group recognition model corresponding to the seed user set.

[0128] In some embodiments, the profile data corresponding to the seed users among multiple initial users can be used as positive samples, and the profile data corresponding to the non-seed users among multiple initial users can be used as negative samples to train a similar population recognition model.

[0129] Step S303: Remove the signaling data from the multiple signaling data sets where the initial user is a seed user, to obtain the remaining user set.

[0130] Step S304: Based on the trained similar user identification model corresponding to the seed user set, the remaining user set is filtered to obtain a second similar user set that is similar to the seed user set.

[0131] In this embodiment of the application, the second set of similar groups identified by machine learning algorithms can divide similar groups from the dimension of user profile data, thereby improving the accuracy of similar group identification in this application.

[0132] Figure 4 This is an optional flowchart illustrating a similar crowd identification method provided in an embodiment of this application. This method can be executed by the processor of a computer device. Based on... Figure 1 The integrated association graph includes a third user vertex, at least one third connection relationship, and a weight sum corresponding to each third connection relationship. The third connection relationship indicates that the initial users corresponding to the two third user vertices have an association relationship. Figure 1 The method further includes steps S401 to S403, which combine... Figure 4 The steps shown are explained.

[0133] Step S401: In the comprehensive association graph, retain the third user vertex corresponding to the seed user in the seed user set and the third user vertex corresponding to the target similar user in the target similar group set.

[0134] Based on the above example, if the target similar user group includes (m, q), and the seed users include n and p, then retain... Figure 2E The third user vertices corresponding to seed user n and seed user p in the comprehensive association graph shown are respectively, and the third user vertices corresponding to the target similar group set (m, q) are respectively.

[0135] Step S402: For each target similar user, sum the weights corresponding to the third connection relationship between the third user vertex corresponding to the target similar user and the third user vertex corresponding to the seed user to obtain the similarity between the target similar user and the seed user set.

[0136] Based on the above example, for a target similar user m, the target similar user m has a third connection relationship "mn" with a weight sum of 3 and a third connection relationship "mp" with a weight sum of 1. After summing, the similarity between the target similar user m and the seed user set is 4. For a target similar user n, the target similar user q has a third connection relationship "qn" with a weight sum of 1. After summing, the similarity between the target similar user q and the seed user set is 1.

[0137] Step S403: Based on the similarity corresponding to each target similar user, generate a target similar group set with ranking information.

[0138] Based on the above example, the resulting target similar user set with ranking information includes (m, q), where the similarity of target similar user m is 4 and the similarity of target similar user q is 1. Target similar user m is located before target similar user q, that is, target similar user m is more similar to the seed user set than target similar user q.

[0139] In this embodiment, by summing the weights of the relationships between each target similar user and the seed user in the comprehensive association graph, the overall relationship between each target similar user and various sub-users can be accurately obtained, thereby determining the similarity between the target similar user and the seed user set. At the same time, based on the similarity corresponding to each target similar user, a target similar group set with ranking information is generated, which can obtain richer identification information on the basis of obtaining the target similar group, providing a data foundation for subsequent actual implementation scenarios.

[0140] The following describes the application of the similar group identification method provided in this disclosure in real-world scenarios.

[0141] Please see Figure 5 This is an optional flowchart illustrating a similar crowd identification method provided in an embodiment of this application. This method can be executed by the processor of a computer device and will combine... Figure 5 The steps shown are explained.

[0142] Step S501: Based on location signaling data, use rasterization processing and graph model construction to calculate the similar population dataset A;

[0143] Here, the similar population dataset A corresponds to the first similar population set in the above embodiment.

[0144] Step S502: Use look-alike algorithms to calculate the regional seed user similarity dataset B.

[0145] Here, the similar population dataset B corresponds to the second similar population set in the above embodiment.

[0146] Step S503: Find the intersection of dataset A and dataset B to obtain dataset C. Rank dataset C based on similarity to obtain the ranked similar groups of people.

[0147] In this step, the intersection of dataset A and dataset B is obtained, resulting in dataset C = A∩B. The ranked similar groups can be denoted as Cs. The ranked similar groups correspond to the target similar group set with ranking information in the above embodiments.

[0148] In some embodiments, it can be achieved through Figure 6 The illustrated example calculates a similar population dataset A. Please refer to [link / reference]. Figure 6 This method can be executed by the processor of a computer device, combining Figure 6 The steps shown are explained.

[0149] Step S601: Obtain the location signaling data set D within the set time range and the set area range.

[0150] Among them, the location signaling data set D corresponds to multiple signaling data in the above embodiments, and this data set must contain at least the data fields shown in Table 1:

[0151] Table 1

[0152] IMSI International Mobile Subscriber Identity MSISDN User's mobile phone number stationId Base station unique identifier countyId The administrative district / county code of the base station latitude Base station latitude longitude Base station longitude timestamp Signaling generation timestamp

[0153] Step S602: Based on the Google S2 algorithm, select multiple levels and perform rasterization processing on the signaling data in step S601. The rasterization levels are {S1, S2...S...} n The processed datasets are {D1, D2...D}. n}

[0154] Please see Figure 7 It illustrates a multi-level raster transformation. With a raster processing level of S1, signaling dataset D1 can be obtained; with a raster processing level of S2, signaling dataset D2 can be obtained; and so on, with a raster processing level of S... n In the case of a multi-level raster, the signaling dataset D can be obtained. n Among them, the signaling dataset D i The data set formed by the grid area corresponding to each signaling data in the above embodiments at the level.

[0155] Step S603: Construct a bipartite graph for the dataset {D1, D2...D...} in step S602. nUsing the mobile phone user's unique identifier IMSI and the base station grid ID as the vertices of the graph, respectively, construct a bipartite graph {T1, T2, ..., T}. n}

[0156] For dataset D i (i∈[1,n]), with dataset D i All IMSI mobile phone users and grid IDs appearing in the dataset are used as vertices of the bipartite graph (corresponding to the first user vertex and grid vertex in the above embodiment, respectively), and the vertices of the constructed bipartite graph are connected based on the signaling data in the dataset that has been rasterized (corresponding to the first connection relationship between the grid vertex and the first user vertex in the above embodiment).

[0157] For example, please refer to the signaling data and corresponding bipartite graph connection diagram shown in Table 2. Figure 8 .

[0158] Table 2

[0159] m 188xxxxxxxx a m 188xxxxxxxx c n 139xxxxxxxx b n 139xxxxxxxx c

[0160] For mobile user m in Table 2, mobile user m has associated records with grids a and c, therefore mobile user m is connected to grids a and c by edges. For mobile user n in Table 2, mobile user n has associated records with grids b and c, therefore mobile user n is connected to grids b and c by edges. Therefore, we can obtain the following... Figure 8 The bipartite graph.

[0161] Step S604: Based on the bipartite graph set {T1, T2...T...} calculated in step S603... n Construct a multi-layer graph {G1, G2, ..., G} with mobile users as vertices. n}

[0162] For a single bipartite graph T i (i∈[1,n]), the corresponding generated graph G i The vertex is the dataset D in step S602. i All IMSI mobile phone users, Figure G i The edge connection lines are processed according to steps S6041 to S6044.

[0163] Step S6041: Traverse the bipartite graph T i A list of vertices for all raster IDs {V D1 V D2 ...V DL}

[0164] Where L is the number of raster IDs.

[0165] Step S6042, for the bipartite graph T i A single raster ID vertex V Dj Reading and V Dj List of all connected IMSI mobile user vertices {V I1 V I2 ...V IN}, where N is related to V Dj The number of connected mobile user vertices, where V Ii The second user vertex in the initial user association graph corresponding to the i-th level in the above embodiment.

[0166] Step S6043, for {V I1 V I2 ...V IN}, in graph G i In the middle, {V I1 V I2 ...V IN Connect each pair of mobile user vertices in the array.

[0167] Step S6044: For all grid ID vertices in the bipartite graph, repeat steps S6042 and S6043 above to construct graph G. i All edges.

[0168] Among them, Figure G i All edges correspond to the second connection relationship between the second user vertices corresponding to each of the two initial users in the above embodiment.

[0169] Please see Figure 9 The diagram shows the relationship between merged mobile phone users. Figure 9 In the bipartite graph, mobile users m and n are both connected to base station grid c by edges. Figure 9 In the diagram, m and n are also connected by an edge.

[0170] Step S605: For the multi-layer graph set {G1, G2...G...} obtained in step S604... n}, merge the vertices to form graph G.

[0171] Here, graph G corresponds to the comprehensive association graph in the above embodiments. The vertices in graph G are a multi-layer graph set {G1, G2...G...} n The set of all vertices in graph G, and the set of edges in graph G is a multi-level graph set {G1, G2, ..., G}. n The set of all edges in graph G, and the logic for calculating the weights of edges in graph G are as follows:

[0172] Because of the multi-layer graph set {G1,G2...G... nThe multi-layer graph G is calculated from location signaling data based on different grid levels. Different grid levels represent the relationships between users within different grid size ranges. Therefore, for different grid levels, the generated multi-layer graph G... i Set a weight value Q i The smaller the range represented by the grid level, the greater its weight value.

[0173] For the weight value E of the edge in the merged graph G Q If it is in the multi-level graph set {G1, G2...G...} n If} appears, its weight value will increase by 1*Q accordingly. i E Q The value is equal to its value in all occurrences of G. i The corresponding weight Q i The sum of multiplied by 1.

[0174] Step S606: Based on the merged graph G generated in step S605, remove all vertices in graph G that exist in set M to obtain the seed user similar group graph G. m The user set in the seed user similarity group graph is the similarity group dataset A.

[0175] In some embodiments, it can be achieved through Figure 10 The illustrated embodiment calculates a similar population dataset B. Please refer to [link / reference]. Figure 10 This method can be executed by the processor of a computer device, combining Figure 10 The steps shown are explained.

[0176] Step S1001: Obtain the location signaling data set D within the set time range and the set area range.

[0177] The dataset must contain at least the data fields shown in Table 1.

[0178] Step S1002: Obtain the profile data P of all mobile phone users in the data set D in step S1001.

[0179] The description of the P field in the portrait dataset is shown in Table 3 below:

[0180] Table 3

[0181] IMSI International Mobile Subscriber Identity SEX gender AGE age MAGE_STATUS Marital status EDU Education OCPN Profession USER_STAR User rating TERM_BRAND Terminal Brands TERM_MDL Terminal Model APP_PRE APP preferences Other portrait features

[0182] Step S1003: Using the portrait dataset P from step S1002, train the seed user model based on the look-alike machine learning algorithm to obtain the training model.

[0183] Step S1004: Using the dataset D from step S1002 and the seed user set M, perform a difference operation based on the IMSI field to obtain the set of users D remaining in dataset D after removing the seed users. r The trained model from step S1003 is used on dataset D. r The model is used to predict and obtain a set of seed users similar to the target user group, B.

[0184] In some embodiments, it can be achieved through Figure 11 The illustrated embodiment sorts the populations in the similar population set C based on the weights of the adjacent edges of vertices in graph Gm. Please refer to [link to relevant documentation]. Figure 11 This method can be executed by the processor of a computer device, combining Figure 11 The steps shown are explained.

[0185] Step S1101: Obtain the sorting object, which is all mobile phone users {C1C2...C1} in the similar user group set C. n};

[0186] Step S1102, for a single C i In graph Gm, find all its edges, sum the weights of all edges whose other vertex is the seed user, and use this sum as the weight of that vertex C. i weight V Q Represents vertex C i Similarity to seed users.

[0187] Step S1103, according to V in step S1102 Q Sort from largest to smallest to obtain the sorted similar population C. s C s This is the final set of similar population results.

[0188] This application proposes a method for identifying similar groups of people in a region based on multi-layer grids and multi-layer graph models. This method solves the problem in the aforementioned prior art solution 1 where user association at a single base station dimension leads to different users being actually close but not effectively identified because they do not have signaling data under the same base station. By using multi-layer grids, compared with a single base station, it achieves personnel identification in multiple ranges. At the same time, through the construction and merging of multi-layer graph models, it effectively identifies the closeness of relationships caused by the length of time spent at the base station in related solutions. The longer two users stay in similar locations, the more associations they have in the multi-layer grid, and the corresponding multi-layer connected graph will have connecting edges. According to the merging method of multi-layer connected graphs, the edge weight of the two users after merging is larger, and the calculated similarity is higher.

[0189] Furthermore, compared to similar groups identified by simple look-alike machine learning algorithms, this application fully utilizes the advantages of location signaling data, fully explores the spatiotemporal relationships of location signaling, and calculates similar groups through multi-layer grid and multi-layer connected graph models. These similar groups are then fused with those calculated by machine learning algorithms. Based on the characteristics of location signaling, groups of people within a spatiotemporal range have more similar attribute features. The fusion method proposed in this application improves the accuracy of identifying similar groups and also enhances the density of the identified similar groups.

[0190] Based on the foregoing embodiments, this application provides a similar crowd identification device. The device includes various units and modules included in each unit, which can be implemented by a processor in a computer device; of course, it can also be implemented by specific logic circuits. In the implementation process, the processor can be a central processing unit (CPU), a microprocessor unit (MPU), a digital signal processor (DSP), or a field programmable gate array (FPGA), etc.

[0191] Figure 12 This is a schematic diagram of the composition structure of a similar crowd identification device provided in an embodiment of this application, as shown below. Figure 12 As shown, the similar crowd identification device 1200 includes: an acquisition module 1210, a first similar crowd determination module 1220, a second similar crowd determination module 1230, and a target similar crowd determination module 1240, wherein:

[0192] The acquisition module 1210 is used to acquire multiple signaling data and signaling attributes of each signaling data; the signaling attributes include location information and the initial user corresponding to the signaling data.

[0193] The first similar group determination module 1220 is used to perform multi-level rasterization processing on the multiple signaling data based on the location information of each signaling data to obtain the correlation information between each initial user, and to determine the first similar group set similar to the seed user set.

[0194] The second similar group determination module 1230 is used to filter multiple initial users corresponding to the multiple signaling data based on the trained similar group identification model corresponding to the seed user set, and obtain a second similar group set similar to the seed user set.

[0195] The target similar population determination module 1240 is used to take the intersection of the first similar population set and the second similar population set as the target similar population set.

[0196] In some embodiments, the first similar population determination module 1220 is further configured to:

[0197] Based on the location information of each signaling data, the plurality of signaling data are subjected to multiple levels of rasterization processing to obtain the raster division result corresponding to each level; the raster division result includes the raster area corresponding to each signaling data at the level;

[0198] For each level, based on the initial user and grid area corresponding to each signaling data, an initial user relationship graph corresponding to that level is determined; the initial user relationship graph represents whether there is a relationship between each initial user;

[0199] A comprehensive association graph is generated based on the level weight corresponding to each level and the initial user association graph corresponding to each level; the comprehensive association graph includes the association degree information between each initial user.

[0200] Based on the correlation information between the initial users in the comprehensive correlation graph, abnormal users are removed from the multiple initial users to obtain a first similar group set that is similar to the seed user set.

[0201] In some embodiments, the grid division result includes the grid region corresponding to each signaling data at the level; the first similar population determination module 1220 is further configured to:

[0202] For each level of rasterization processing, obtain the set of raster regions corresponding to that level;

[0203] Based on the location information of each signaling data and each grid region in the grid region set, the grid region corresponding to each signaling data at the level is determined.

[0204] In some embodiments, the first similar population determination module 1220 is further configured to:

[0205] Based on the association between the initial user and the grid region in each of the signaling data, a bipartite graph is constructed; the bipartite graph represents whether there is an association between each initial user and each grid region;

[0206] The bipartite graph is converted into the initial user association graph corresponding to the level.

[0207] In some embodiments, the first similar population determination module 1220 is further configured to:

[0208] Each of the grid regions is taken as a grid vertex in the bipartite graph, and each of the initial users is taken as the first user vertex in the bipartite graph;

[0209] For each of the signaling data, a first connection relationship is established between the grid vertex corresponding to the signaling data and the first user vertex to obtain the bipartite graph; the first connection relationship indicates that there is an association between the grid region corresponding to the grid vertex and the initial user corresponding to the first user vertex.

[0210] In some embodiments, the first similar population determination module 1220 is further configured to:

[0211] For each grid vertex in the bipartite graph, an initial user set corresponding to the grid vertex is obtained in the bipartite graph; the initial user set includes the initial users corresponding to all first user vertices that have a first connection relationship with the grid vertex;

[0212] Each of the initial users is designated as the second user vertex in the initial user association graph;

[0213] For each grid vertex corresponding to the initial user set, a second connection relationship is established between the second user vertices corresponding to every two initial users in the initial user set, resulting in an initial user association graph corresponding to the level; the second connection relationship indicates that the initial users corresponding to the two second user vertices have an association relationship.

[0214] In some embodiments, the initial user association graph includes second user vertices and at least one second connection relationship, wherein the second connection relationship indicates that the initial users corresponding to two corresponding second user vertices are associated; the first similar user group determination module 1220 is further configured to:

[0215] Based on the union of the second user vertices in the initial user association graph corresponding to each level, the third user vertex in the comprehensive association graph is constructed.

[0216] Based on the union of at least one second connection relationship in the initial user association graph corresponding to each level, at least one third connection relationship is obtained in the comprehensive association graph; the third connection relationship indicates that the initial users corresponding to the two third user vertices have an association relationship.

[0217] Based on the level weight corresponding to each level and at least one second connection relationship in the initial user association graph corresponding to each level, the weight sum corresponding to each third connection relationship in the comprehensive association graph is determined; the weight sum represents the association degree information between the two initial users.

[0218] In some embodiments, the abnormal users include other users and seed users in the seed user set; the first similar user group determination module 1220 is further configured to:

[0219] Based on the correlation information between the initial users in the comprehensive correlation graph, the initial users who have no connection relationship with each of the seed users are selected as the other users among the plurality of initial users;

[0220] The other users and the seed users are removed from the plurality of initial users to obtain a first similarity group set that is similar to the seed user set.

[0221] In some embodiments, the comprehensive association graph includes a third user vertex, at least one third connection relationship, and a weight sum corresponding to each third connection relationship, wherein the third connection relationship indicates that the initial users corresponding to two third user vertices are associated; the first similar user group determination module 1220 is further configured to:

[0222] The comprehensive association graph retains the third user vertex corresponding to the seed user in the seed user set and the third user vertex corresponding to the target similar user in the target similar group set;

[0223] For each target similar user, the weights corresponding to the third connection relationship between the third user vertex corresponding to the target similar user and the third user vertex corresponding to the seed user are summed to obtain the similarity between the target similar user and the set of seed users.

[0224] Based on the similarity of each target similar user, a target similar group set with ranking information is generated.

[0225] In some embodiments, the second similar population determination module 1230 is further configured to:

[0226] Obtain the profile data corresponding to each of the initial users corresponding to the plurality of signaling data, as a profile dataset;

[0227] Based on the aforementioned profile dataset, a model is trained to obtain a similar population recognition model corresponding to the seed user set.

[0228] Remove the signaling data from the multiple signaling data sets where the initial user is a seed user, to obtain the remaining user set;

[0229] Based on the trained similar user identification model corresponding to the seed user set, the remaining user set is filtered to obtain a second similar user set that is similar to the seed user set.

[0230] The descriptions of the apparatus embodiments above are similar to those of the method embodiments above, and have similar beneficial effects. In some embodiments, the functions or modules included in the apparatus provided in this application can be used to perform the methods described in the method embodiments above. For technical details not disclosed in the apparatus embodiments of this application, please refer to the descriptions of the method embodiments of this application for understanding.

[0231] It should be noted that, in the embodiments of this application, if the above-mentioned similar group identification method is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiments of this application, or the part that contributes to the related technology, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, mobile hard drives, read-only memory (ROM), magnetic disks, or optical disks. Thus, the embodiments of this application are not limited to any specific hardware, software, or firmware, or any combination of hardware, software, and firmware.

[0232] This application provides a computer device including a memory and a processor. The memory stores a computer program that can run on the processor. When the processor executes the program, it implements some or all of the steps in the above-described method.

[0233] This application provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements some or all of the steps in the above-described method. The computer-readable storage medium can be transient or non-transient.

[0234] This application provides a computer program including computer-readable code, wherein when the computer-readable code is executed in a computer device, a processor in the computer device performs some or all of the steps in the above-described method.

[0235] This application provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program. When the computer program is read and executed by a computer, it implements some or all of the steps in the above-described method. This computer program product can be implemented specifically through hardware, software, or a combination thereof. In some embodiments, the computer program product is specifically embodied as a computer storage medium; in other embodiments, the computer program product is specifically embodied as a software product, such as a software development kit (SDK), etc.

[0236] It should be noted that the descriptions of the various embodiments above tend to emphasize the differences between them, while their similarities or commonalities can be referred to interchangeably. The descriptions of the above embodiments of the device, storage medium, computer program, and computer program product are similar to the descriptions of the above method embodiments and have similar beneficial effects. For technical details not disclosed in the embodiments of the device, storage medium, computer program, and computer program product of this application, please refer to the descriptions of the method embodiments of this application for understanding.

[0237] Figure 13 This application provides a hardware entity diagram of a computer device as an embodiment of the present application, such as... Figure 13 As shown, the hardware entity of the computer device 1300 includes a processor 1301 and a memory 1302, wherein the memory 1302 stores a computer program that can run on the processor 1301, and the processor 1301 executes the program to implement the steps in the method of any of the above embodiments.

[0238] The memory 1302 stores computer programs that can run on the processor. The memory 1302 is configured to store instructions and applications that can be executed by the processor 1301. It can also cache data to be processed or already processed (e.g., image data, audio data, voice communication data, and video communication data) in the processor 1301 and various modules in the computer device 1300. It can be implemented by flash memory or random access memory (RAM).

[0239] When processor 1301 executes a program, it implements the steps of any of the above-mentioned similar crowd identification methods. Processor 1301 typically controls the overall operation of computer device 1300.

[0240] This application provides a computer storage medium storing one or more programs that can be executed by one or more processors to implement the steps of the similar crowd identification method as described in any of the above embodiments.

[0241] It should be noted that the descriptions of the storage medium and device embodiments above are similar to the descriptions of the method embodiments above, and have similar beneficial effects. For technical details not disclosed in the storage medium and device embodiments of this application, please refer to the descriptions of the method embodiments of this application for understanding.

[0242] The aforementioned processor can be at least one of the following: Application Specific Integrated Circuit (ASIC), Digital Signal Processor (DSP), Digital Signal Processing Device (DSPD), Programmable Logic Device (PLD), Field Programmable Gate Array (FPGA), Central Processing Unit (CPU), Controller, Microcontroller, and Microprocessor. It is understood that other electronic devices can also implement the functions of the aforementioned processor, and this application does not specifically limit the specific implementation.

[0243] The aforementioned computer storage media / memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), magnetic random access memory (FRAM), flash memory, magnetic surface memory, optical disc, or compact disc read-only memory (CD-ROM), etc.; or it can be various terminals that include one or any combination of the above-mentioned memories, such as mobile phones, computers, tablet devices, personal digital assistants, etc.

[0244] It should be understood that the phrase "one embodiment" or "an embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of this application. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. It should be understood that in the various embodiments of this application, the sequence numbers of the above steps / processes do not imply a sequential order of execution; the execution order of each step / process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. The sequence numbers of the above embodiments of this application are merely descriptive and do not represent the superiority or inferiority of the embodiments.

[0245] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0246] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0247] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0248] In addition, each functional unit in the various embodiments of this application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0249] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media that can store program code, such as mobile storage devices, read-only memory (ROM), magnetic disks, or optical disks.

[0250] Alternatively, if the integrated units described above are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence or the part that contributes to related technologies, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, magnetic disks, or optical disks.

[0251] The above description is merely an embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method of similar people identification, characterized by, The method includes: Obtain multiple signaling data and signaling attributes for each signaling data; the signaling attributes include location information and the initial user corresponding to the signaling data; Based on the location information of each signaling data, the multiple signaling data are subjected to multiple levels of rasterization processing to obtain the correlation information between each initial user, and a first similar group set similar to the seed user set is determined; the correlation information is generated based on the raster division results of each level and is used to characterize whether there is a correlation between the corresponding two initial users. Based on the trained similar user identification model corresponding to the seed user set, multiple initial users corresponding to the multiple signaling data are screened to obtain a second similar user set similar to the seed user set; The intersection of the first set of similar people and the second set of similar people is taken as the target set of similar people.

2. The method of claim 1, wherein, The step of performing multi-level rasterization processing on the multiple signaling data based on the location information of each signaling data to obtain the correlation information between the initial users and to determine a first similar group set similar to the seed user set includes: Based on the location information of each signaling data, the plurality of signaling data are subjected to multiple levels of rasterization processing to obtain the raster division result corresponding to each level; the raster division result includes the raster area corresponding to each signaling data at the level; For each level, based on the initial user and grid area corresponding to each signaling data, an initial user relationship graph corresponding to that level is determined; the initial user relationship graph represents whether there is a relationship between each initial user; A comprehensive association graph is generated based on the level weight corresponding to each level and the initial user association graph corresponding to each level; the comprehensive association graph includes the association degree information between each initial user. Based on the correlation information between the initial users in the comprehensive correlation graph, abnormal users are removed from the multiple initial users to obtain a first similar user set similar to the seed user set; the abnormal users include other users and seed users in the seed user set; the other users are initial users who have no connection relationship with the seed users.

3. The method of claim 2, wherein, The raster division result includes the raster region corresponding to each signaling data at the level; the process of performing rasterization processing on the multiple signaling data at multiple levels based on the location information of each signaling data to obtain the raster division result corresponding to each level includes: For each level of rasterization processing, obtain the set of raster regions corresponding to that level; Based on the location information of each signaling data and each grid region in the grid region set, the grid region corresponding to each signaling data at the level is determined.

4. The method of claim 2, wherein, The step of determining the initial user association graph corresponding to the level based on the initial user and grid area corresponding to each signaling data includes: Based on the association between the initial user and the grid region in each of the signaling data, a bipartite graph is constructed; the bipartite graph represents whether there is an association between each initial user and each grid region; The bipartite graph is converted into the initial user association graph corresponding to the level.

5. The method according to claim 4, characterized in that, The construction of a bipartite graph based on the association between the initial user and the grid region in each of the signaling data includes: Each of the grid regions is taken as a grid vertex in the bipartite graph, and each of the initial users is taken as the first user vertex in the bipartite graph; For each of the signaling data, a first connection relationship is established between the grid vertex corresponding to the signaling data and the first user vertex to obtain the bipartite graph; the first connection relationship indicates that there is an association between the grid region corresponding to the grid vertex and the initial user corresponding to the first user vertex.

6. The method according to claim 5, characterized in that, The step of converting the bipartite graph into the initial user association graph corresponding to the level includes: For each grid vertex in the bipartite graph, an initial user set corresponding to the grid vertex is obtained in the bipartite graph; the initial user set includes the initial users corresponding to all first user vertices that have a first connection relationship with the grid vertex; Each of the initial users is designated as the second user vertex in the initial user association graph; For each grid vertex corresponding to the initial user set, a second connection relationship is established between the second user vertices corresponding to every two initial users in the initial user set, resulting in an initial user association graph corresponding to the level; the second connection relationship indicates that the initial users corresponding to the two second user vertices have an association relationship.

7. The method according to claim 2, characterized in that, The initial user association graph includes second user vertices and at least one second connection relationship, wherein the second connection relationship indicates that the initial users corresponding to the two second user vertices are associated with each other; The process of generating a comprehensive association graph based on the level weight corresponding to each level and the initial user association graph corresponding to each level includes: Based on the union of the second user vertices in the initial user association graph corresponding to each level, the third user vertex in the comprehensive association graph is constructed. Based on the union of at least one second connection relationship in the initial user association graph corresponding to each level, at least one third connection relationship is obtained in the comprehensive association graph; the third connection relationship indicates that the initial users corresponding to the two third user vertices have an association relationship. Based on the level weight corresponding to each level and at least one second connection relationship in the initial user association graph corresponding to each level, the weight sum corresponding to each third connection relationship in the comprehensive association graph is determined; the weight sum represents the association degree information between the two initial users.

8. The method according to claim 2, characterized in that, The comprehensive association graph includes third user vertices, at least one third connection relationship, and a weight sum corresponding to each third connection relationship, wherein the third connection relationship indicates that the initial users corresponding to two third user vertices are associated; the method further includes: The comprehensive association graph retains the third user vertex corresponding to the seed user in the seed user set and the third user vertex corresponding to the target similar user in the target similar group set; For each target similar user, the weights corresponding to the third connection relationship between the third user vertex corresponding to the target similar user and the third user vertex corresponding to the seed user are summed to obtain the similarity between the target similar user and the set of seed users. Based on the similarity of each target similar user, a target similar group set with ranking information is generated.

9. A similar crowd identification device, characterized in that, The device includes: An acquisition module is used to acquire multiple signaling data and signaling attributes of each signaling data; the signaling attributes include location information and the initial user corresponding to the signaling data; The first similar user group determination module is used to perform multiple levels of rasterization processing on the multiple signaling data based on the location information of each signaling data to obtain the correlation information between each initial user and to determine the first similar user group set similar to the seed user set; the correlation information is generated based on the rasterization results of each level and is used to characterize whether there is a correlation between the corresponding two initial users. The second similar user group determination module is used to filter multiple initial users corresponding to the multiple signaling data based on the trained similar user group identification model corresponding to the seed user set, and obtain a second similar user group set similar to the seed user set. The target similar population determination module is used to take the intersection of the first similar population set and the second similar population set as the target similar population set.

10. A computer device comprising a memory and a processor, the memory storing a computer program executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method according to any one of claims 1 to 8.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the method according to any one of claims 1 to 8.

Images