User association strength identification method and device, equipment, medium and product

Abstract

The invention provides a user association strength identification method and device, equipment, a medium and a product, and relates to the technical field of big data, the method comprises the following steps: obtaining first information of a plurality of users, the first information of each user comprising at least one of user basic information, communication data and position data; determining an association feature value between every two users according to the first information; constructing a social network of the plurality of users according to the associated feature values; and based on the social network, determining the comprehensive association strength of a first to-be-identified user and a second to-be-identified user, the first to-be-identified user and the second to-be-identified user being users in the plurality of users. Thus, the association strength between the two users is determined based on the social network among the multiple users, and the problem that the recognition accuracy is not high when user relationship recognition is carried out only based on the data corresponding to the two users to be recognized is solved.

Description

Technical Field

[0001] The present application relates to the field of big data technology, and in particular to a method, device, equipment, medium and product for identifying user association strength. Background Art

[0002] Related technologies have proposed methods for determining whether users have a family relationship based on their communication and location relationships. Alternatively, methods for identifying family and friend relationships based on user profile information, call behavior, and location behavior can be used. Alternatively, methods for determining whether users have a residential relationship based on the similarity of communication relationships and communication environments based on user call data can be used. However, these solutions all rely on the relevant data of the two users to be identified, resulting in low recognition accuracy. Summary of the Invention

[0003] The embodiments of the present application provide a method, apparatus, device, medium and product for identifying user association strength, which solves the problem of low recognition accuracy in current user relationship recognition based only on data corresponding to the user to be identified.

[0004] In a first aspect, to achieve the above-mentioned objectives, embodiments of the present application provide a method for identifying user association strength, comprising:

[0005] Acquire first information of a plurality of users, where the first information of each user includes at least one of user basic information, communication data, and location data;

[0006] Determining, based on the first information, a correlation feature value between every two users;

[0007] constructing a social network of a plurality of the users according to the associated feature values;

[0008] Based on the social network, a comprehensive correlation strength between a first user to be identified and a second user to be identified is determined, wherein the first user to be identified and the second user to be identified are users among the plurality of users.

[0009] Determining the correlation feature value between every two users based on the first information includes:

[0010] Determining, based on the first information, at least one characteristic value representing an association relationship between two users;

[0011] determining the associated feature value based on the at least one feature value;

[0012] The at least one characteristic value includes at least one of the following:

[0013] a first feature value determined according to the user basic information, the first feature value being used to characterize a factual association degree between two users;

[0014] a second characteristic value determined according to the communication data, the second characteristic value being used to characterize the degree of association between the social circles of the two users;

[0015] A third eigenvalue is determined according to the location data, where the third eigenvalue is used to characterize a location association degree between the two users.

[0016] The basic user information includes at least one of user identity information, subscription package information and payment information;

[0017] Determining, based on the first information, at least one characteristic value representing an association relationship between two users includes:

[0018] According to the user basic information, if two users meet a first condition, determining that the first characteristic values ​​corresponding to the two users are a first value, wherein the first condition includes at least one of the following:

[0019] The user identity information bound to the two users is the same;

[0020] The subscription package indicates that the two users are family accounts;

[0021] The subscription package indicates the relationship between the two users as primary and secondary cards;

[0022] The payment information indicates that there is an agency collection and payment relationship between the two users.

[0023] The communication data includes the total number of calls, total call duration and total number of text messages between the user and different users;

[0024] Determining, based on the first information, at least one characteristic value representing an association relationship between two users includes:

[0025] Normalizing the total number of calls, the total duration of calls, and the total number of text messages in the communication data;

[0026] The normalized total number of calls, the total call duration, and the total number of text messages are weightedly summed to obtain the second eigenvalue.

[0027] Determining, based on the first information, at least one characteristic value representing the association relationship between two users includes:

[0028] Determining consistency between the permanent locations of the two users based on the location data;

[0029] obtaining a permanent location characteristic value corresponding to the consistency of the permanent location;

[0030] Determining, based on the location data, a characteristic value of a degree of overlap of the location trajectories of the two users;

[0031] A weighted sum is performed on the permanent location eigenvalue and the position trajectory coincidence eigenvalue to obtain the third eigenvalue.

[0032] The step of determining the associated characteristic value according to the at least one characteristic value includes:

[0033] performing a weighted summation on at least two of the first eigenvalue, the second eigenvalue, and the third eigenvalue to obtain a fourth eigenvalue;

[0034] The fourth eigenvalue is mapped using a nonlinear function to determine the associated eigenvalue.

[0035] The step of constructing a social network of multiple users based on the associated feature values ​​includes:

[0036] An undirected weighted graph is established based on the multiple users and the associated feature values, wherein a vertex of the undirected weighted graph represents the user, and a weight of an edge between two vertices of the undirected weighted graph represents the associated feature value; and the undirected weighted graph represents the social network.

[0037] Determining the comprehensive correlation strength between the first user to be identified and the second user to be identified based on the social network includes:

[0038] In the social network, obtaining multiple paths between the first user to be identified and the second user to be identified;

[0039] Calculate the correlation strength corresponding to each path;

[0040] Determining the weight of the association strength corresponding to each path according to the path length of each path;

[0041] The association strengths corresponding to the various paths are weighted and summed according to the weights to determine the comprehensive association strength.

[0042] In a second aspect, to achieve the above-mentioned purpose, an embodiment of the present application provides a user association strength identification device, comprising:

[0043] An acquisition module, configured to acquire first information of a plurality of users, wherein the first information of each user includes at least one of basic user information, communication data, and location data;

[0044] A first determining module, configured to determine an association feature value between every two users based on the first information;

[0045] A construction module, configured to construct a social network of a plurality of the users according to the associated feature values;

[0046] The second determining module is configured to determine, based on the social network, a comprehensive correlation strength between a first user to be identified and a second user to be identified, wherein the first user to be identified and the second user to be identified are users among the plurality of users.

[0047] In the third aspect, in order to achieve the above-mentioned purpose, an embodiment of the present application provides a user association strength identification device, including a transceiver, a processor, a memory, and a program stored on the memory and runnable on the processor; the transceiver is used to receive and send data under the control of the processor, and the processor implements the user association strength identification method described in the first aspect when executing the program.

[0048] In a fourth aspect, in order to achieve the above-mentioned purpose, an embodiment of the present application provides a readable storage medium on which a program or instruction is stored, and when the program or instruction is executed by a processor, the user association strength identification method as described in the first aspect is implemented.

[0049] In a fifth aspect, in order to achieve the above-mentioned purpose, an embodiment of the present application provides a computer program product, including computer instructions, which, when executed by a processor, implement the user association strength identification method as described in the first aspect.

[0050] The beneficial effects of the above technical solution of this application are as follows:

[0051] In an embodiment of the present application, first, first information of multiple users is obtained, where the first information of each user includes at least one of basic user information, communication data, and location data. Second, based on the first information, the association feature value between each two users is determined. Third, based on the association feature value, a social network of the multiple users is constructed. Finally, based on the social network, the comprehensive association strength between the first user to be identified and the second user to be identified is determined, where the first user to be identified and the second user to be identified are users among the multiple users. In this way, the association strength between two users to be identified is determined based on the social network between multiple users, which improves the accuracy of user association strength identification and solves the current problem of low accuracy in determining the association strength between users based solely on the relevant data of the two users to be identified. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 A flowchart of a method for identifying user association strength according to an embodiment of the present application is shown;

[0053] Figure 2A schematic diagram of an undirected right graph service structure according to an embodiment of the present application;

[0054] Figure 3 A schematic diagram of an undirected weighted graph data structure according to an embodiment of the present application;

[0055] Figure 4 This is a schematic diagram of the structure of a device for identifying user association strength according to an embodiment of the present application;

[0056] Figure 5 This is a structural diagram of a user association strength identification device according to an embodiment of the present application. DETAILED DESCRIPTION

[0057] In order to make the technical problems, technical solutions and advantages to be solved by this application clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0058] It should be understood that references throughout this specification to "one embodiment" or "an embodiment" mean that a particular feature, structure, or characteristic associated with the embodiment is included in at least one embodiment of the present application. Therefore, the appearances of "in one embodiment" or "in an embodiment" throughout this specification do not necessarily refer to the same embodiment. Furthermore, these particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0059] In the various embodiments of the present application, it should be understood that the size of the serial numbers of the following processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0060] Additionally, the terms "system" and "network" are often used interchangeably herein.

[0061] In the embodiments provided herein, it should be understood that "B corresponding to A" means that B is associated with A and B can be determined based on A. However, it should also be understood that determining B based on A does not mean determining B based solely on A; B can also be determined based on A and / or other information.

[0062] Before describing the embodiments of the present application, the relevant technical points are first described:

[0063] Graph computing: Graph computing is a process that uses graphs as data models to represent and solve problems. It is particularly important in the era of big data. Because data is interconnected, graph computing can integrate data from different sources and types into a single graph for analysis, yielding insights that would otherwise be difficult to discover through independent analysis. Graph computing has a wide range of applications in scenarios such as personalized recommendations, social network analysis, knowledge graph construction, e-commerce and financial risk management, network security, and information technology (IT) equipment operations and maintenance.

[0064] The embodiment of the present application provides a method for identifying user association strength, such as Figure 1 As shown, the method includes:

[0065] Step 101: Acquire first information of a plurality of users, where the first information of each user includes at least one of user basic information, communication data, and location data.

[0066] Step 102: Determine the association characteristic value between each two users based on the first information; wherein, this step determines the association characteristic value between the two users based on the first information, so that the determined association characteristic value is related to one or more of the user basic information, communication data and location data. Compared with the method of determining the association between two users based only on call behavior, the association characteristic value obtained in this step is more accurate.

[0067] Step 103 : constructing a plurality of social networks of the users to be identified based on the associated feature values; wherein the social networks include social relationships between a user and one or more other users.

[0068] Step 104 determines the comprehensive strength of association between the first user to be identified and the second user to be identified based on the social network, where the first user to be identified and the second user to be identified are users among the plurality of users. In this step, the comprehensive strength of association between the two users to be identified is determined based on the social network, not only by considering the association feature values ​​between the two users to be identified, but also by considering the association feature values ​​between the two users to be identified and other users in the social network. Compared to calculating the strength of association based solely on data between the two users to be identified, the comprehensive strength of association determined in this step is more accurate.

[0069] In an embodiment of the present application, first, first information of multiple users is obtained, where the first information of each user includes at least one of basic user information, communication data, and location data. Second, based on the first information, the association feature value between each two users is determined. Third, based on the association feature value, a social network of the multiple users is constructed. Finally, based on the social network, the comprehensive association strength between the first user to be identified and the second user to be identified is determined, where the first user to be identified and the second user to be identified are users among the multiple users. In this way, the association strength between two users to be identified is determined based on the social network between multiple users, which improves the accuracy of user association strength identification and solves the current problem of low accuracy in determining the association strength between users based solely on the relevant data of the two users to be identified.

[0070] As an optional implementation, step 102 includes:

[0071] Determining, based on the first information, at least one characteristic value representing an association relationship between two users;

[0072] determining the associated feature value based on the at least one feature value;

[0073] The at least one characteristic value includes at least one of the following:

[0074] a first feature value determined according to the user basic information, the first feature value being used to characterize a factual association degree between two users;

[0075] a second characteristic value determined according to the communication data, the second characteristic value being used to characterize the degree of association between the social circles of the two users;

[0076] A third eigenvalue is determined according to the location data, where the third eigenvalue is used to characterize a location association degree between the two users.

[0077] That is to say, the above optional implementation method is to determine the characteristic value corresponding to each dimension based on the content of each dimension included in the first information. For example, when the first information includes user basic information, it is necessary to determine the first feature representing the factual correlation between the two users based on the user basic information. When the first information includes communication data, it is necessary to determine the second feature representing the correlation between the social circles of the two users based on the communication data; when the first information includes location data, it is necessary to determine the third feature representing the location correlation between the two users based on the location data; subsequently, the correlation characteristic value between the two users is finally determined based on the determined characteristic values ​​of different dimensions. In this way, the correlation between the two users can be determined based on data of different dimensions. Compared with determining the correlation between the two users based on data of a single dimension, the correlation characteristic value determined by the above optional implementation method is more accurate.

[0078] As a specific implementation, the basic user information includes at least one of user identity information, subscription package information, and payment information. User identity information, for example, includes at least one of the following: the ID number, gender, and age associated with the user's phone number; subscription package information, for example, includes the user's subscription package, including family numbers, primary and secondary cards, etc.; and payment information, for example, includes collection and payment. Based on this first information, at least one feature value representing the association relationship between the two users to be identified is determined, including:

[0079] According to the user basic information, if two users to be identified meet a first condition, determining that the first feature values ​​corresponding to the two users to be identified are a first value, where the first value is, for example, 1, and the first condition includes at least one of the following:

[0080] The user identity information bound to the two users is the same;

[0081] The subscription package indicates that the two users are family accounts;

[0082] The subscription package indicates the relationship between the two users as primary and secondary cards;

[0083] The payment information indicates that the two users have a collection and payment relationship. A collection and payment relationship refers to one user collecting or paying money on behalf of another user. For example, one user pays phone bills for another user.

[0084] Here, we use telephone number / mobile number A and telephone number / mobile number B as examples to illustrate the implementation process of the above specific implementation method. After obtaining the basic user information of mobile number A and mobile number B, we determine whether mobile number A and mobile number B are bound to the user's ID card, whether they are a primary and secondary card relationship, whether they are a family number relationship, or whether they are a collection and payment agency relationship. If any of these conditions are met, mobile number A and mobile number B have the highest association strength, and the first eigenvalue is set to 1. At the same time, data on dimensions such as user gender and age can also support the identification of potential relationships between users, such as child relationships, colleague relationships, and friend relationships.

[0085] Taking mobile phone number A and mobile phone number B as an example, the above specific implementation method can create four features based on the obtained user basic information (user mobile phone number binding ID number information, package subscription, payment method, etc.), including whether mobile phone number A and mobile phone number B are bound to the same ID card, whether they are family numbers, whether they are collection and payment relationship, and whether they are primary and secondary card relationship.

[0086] Among them, the ID card consistency feature, the collection and payment relationship feature, the primary and secondary card relationship feature, and the family number relationship feature are all binary variables. If mobile phone numbers A and B match any of these features, it represents the highest correlation strength between mobile phone numbers A and B. The four features can be fused into a factual correlation feature. The feature fusion formula is as follows:

[0087] X 事实关联 =X 身份证 ∪X 代收代付 ∪X 主副卡 ∪X 亲情号

[0088] Among them, if mobile phone numbers A and B are bound to the same ID number, then X 身份证 is 1, otherwise X 身份证 Is 0; if mobile phone numbers A and B have a collection and payment relationship, then X 代收代付 is 1, otherwise X 代收代付 Is 0; if mobile phone numbers A and B are primary and secondary card relationships, then X 主副卡 is 1, otherwise X 主副卡 Is 0, if mobile phone numbers A and B are family numbers, then X 亲情号 is 1, otherwise X 亲情号 is 0. Therefore, if any of the above features is 1, then X 事实关联 Returns 1, otherwise, returns 0. 事实关联 represents the first eigenvalue.

[0089] As another specific implementation, the communication data includes the total number of calls, total call duration, and total number of text messages between the user and different users; wherein, the communication data can also be, for example, social circle data, including call behavior and text message behavior data between users. Based on the communication data, it is possible to deeply analyze whether there is any intersection between users at the communication level and evaluate the closeness of such intersection. Specifically, the duration and number of times can reveal the frequency and preferences of communication between users. In addition, in-depth mining of this data can also reveal the social network structure of the user, including the size of the social circle, the user's position in the social circle, and the intensity of interaction with other users, thereby providing a comprehensive and in-depth perspective for understanding the user's social behavior.

[0090] Wherein, based on the communication data including the total number of calls, total call duration, and total number of text messages between the user and different users, determining at least one characteristic value representing the association relationship between two users according to the first information includes:

[0091] 1) Normalizing the total number of calls, the total call duration, and the total number of text messages in the communication data. This step can reduce the impact of the dimensions and value ranges of the three characteristic data through normalization (or standardization).

[0092] The total number of calls and the total number of text messages can be processed using the min-max normalization method. The min-max normalization formula is as follows:

[0093]

[0094] Among them, X max Indicates the number of calls / text messages the user has with the user with the most calls / text messages among the total number of calls / text messages the user has with other users, X min Indicates the number of calls / text messages between the user and the user with the least number of calls / text messages among the total number of calls / text messages between the user and other users. X indicates the total number of calls / text messages between the user and another user whose association relationship needs to be determined. norm Indicates the normalized value of X.

[0095] Since the total call duration conforms to the normal distribution, the total call duration can be normalized using the Z-score. The Z-score normalization formula is as follows:

[0096]

[0097] 2) Performing a weighted summation on the normalized total number of calls, the total call duration, and the total number of text messages to obtain the second eigenvalue. This step can be expressed as:

[0098] X 交往圈 =β 通话时长 X 通话时长 +β 通话次数 X 通话次数 +β 短信次数 X 短信次数

[0099] Among them, β 通话时长 , β 通话次数 and β 短信次数 is the weight pre-set based on prior knowledge / experience, X 通话时长 represents the normalized call duration, X 通话次数 represents the normalized number of calls, X 短信次数 represents the normalized number of text messages, X 交往圈 represents the second eigenvalue.

[0100] It's important to note that location data is a crucial component of user behavior analysis, encompassing dimensions such as a user's permanent workplace and residence. By analyzing user workplace and residence data, we can reveal the scope of their daily activities and assess the potential strength of their connections. For example, two users who live close together may have more overlap in their lifestyles and even social circles. This geographic proximity often enhances social connections and interactions between users. Based on this, this application introduces a large-scale travel model, which derives feature variables from user location data using two dimensions: permanent workplace / residence comparison and location trajectory comparison. Comparing the consistency of users' permanent workplaces and residences is intended to determine proximity at the user location level; comparing user location trajectory behaviors further identifies user connections. For example, if users A and B have the same permanent residence, this simply indicates that they live adjacent to each other, potentially indicating they are family members or neighbors. Furthermore, if the location trajectories of users A and B overlap significantly on weekends and holidays, the probability that they are family members is high.

[0101] Based on the above content, as another specific implementation, determining at least one feature value representing the association relationship between two users based on the first information includes:

[0102] 1) According to the location data, determine the consistency of the permanent places of the two users; and obtain the permanent place feature value corresponding to the consistency of the permanent places; wherein, the location data in this step is, for example, user location signaling data, and the implementation process of this step is: first, calculate the user's permanent workplace and permanent residence based on the user location signaling data. The calculation logic is: according to the duration and frequency of the user's area (cellid) during the working time period, the top 5 cellids are comprehensively arranged in descending order, and the top 5 cellids are taken as the user's permanent workplace; similarly, the top 5 cellids of the user during non-working periods (weekdays after work and weekends and holidays) are calculated as the user's permanent residence. Then compare the permanent workplaces and permanent residences of the two users (users are represented by mobile phone numbers). If the permanent workplaces or permanent residences of the two users overlap, 1 is returned; if there is no overlap at all, 0 is returned. The expression is as follows:

[0103]

[0104] Among them, X 工作地 =1 means that the permanent work locations of the two users overlap, X 居住地 =1 means that the permanent residences of the two users overlap, X 工作地 =0 means that the permanent work locations of the two users do not overlap, X 居住地 =0 means that the permanent residences of the two users do not overlap, X 常驻地 Represents the resident characteristic value.

[0105] 2) Determining a location trajectory overlap characteristic value of the two users based on the location data; the location data in this step may include location trajectory data, wherein this step may determine the number of overlapping locations by traversing the location trajectory data, and determine the location trajectory overlap characteristic value based on a ratio of the number of overlapping locations to the total number of locations.

[0106] An implementation example of the above steps is: based on the operator's user location signaling data, select the user coordinate trajectory data during the user's rest period (including after get off work on weekdays, weekends, and holidays). Taking into account the issues of data magnitude and computational overhead, a simple logic can be selected to compare the trajectory overlap. The logic is as follows: First, initialize a counter; second, use HashSet (a set based on a hash table) to store the location trajectory coordinates of one of the users; third, traverse the location trajectory coordinate points of the second user. If they are adjacent, the counter is +1; if they are not adjacent, the counter remains unchanged; finally, after the traversal is completed, the counter value / the total number of coordinates of a certain user can be used to obtain the overlap. The calculation formula for the user location trajectory overlap is as follows:

[0107]

[0108] Here, it should be noted that when traversing the location trajectory coordinate points, the time corresponding to the location trajectory coordinate points needs to be considered. For example, if the location trajectory coordinate points of two users coincide within a close time, it is confirmed that the two users coincide at the location trajectory coordinate point.

[0109] 3) Performing a weighted summation on the permanent location eigenvalue and the location trajectory coincidence eigenvalue to obtain the third eigenvalue; wherein this step can be expressed by the following formula:

[0110] X 位置关联 =β 常驻地 X 常驻地 +β 位置重合度 X 位置重合度

[0111] Among them, β 常驻地 and β 常驻地 is the weight pre-set based on prior knowledge / experience, X 常驻地 represents the resident eigenvalue, X 位置重合度 Indicates the position trajectory coincidence characteristic value, X 位置关联 represents the third eigenvalue.

[0112] It should be noted that, based on the above implementations, before step 102, the method further includes:

[0113] Data cleaning is performed on the first information to ensure and improve data quality. Data cleaning includes handling missing and outliers. For example, the distribution and causes of missing and outlier data are analyzed, and targeted methods are selected for deleting and filling missing records. Proper handling of outlier and missing data ensures data quality and consistency, improves data accuracy and reliability, and enhances the model's predictive power and generalizability.

[0114] As a specific implementation, determining the associated characteristic value according to the at least one characteristic value includes:

[0115] A weighted sum is performed on at least two of the first eigenvalue, the second eigenvalue, and the third eigenvalue to obtain a fourth eigenvalue. For example, in the case where the first information includes basic user information, communication data, and location data, this step is to perform a weighted sum on the first eigenvalue, the second eigenvalue, and the third eigenvalue to obtain the fourth eigenvalue. In this case, this step can be expressed as follows:

[0116] w 关联强度 =β 事实关联 X 事实关联 +β 交往圈关联 X 交往圈关联 +β 位置关联X 位置关联

[0117] Among them, β 事实关联 , β 交往圈关联 and β 位置关联 is the weight pre-set based on prior knowledge / experience, X 事实关联 represents the first eigenvalue, X 交往圈关联 represents the second eigenvalue, X 位置关联 represents the third eigenvalue, w 关联强度 represents the fourth eigenvalue.

[0118] The fourth eigenvalue is mapped using a nonlinear function to determine the associated eigenvalue.

[0119] It should be noted that due to the limitations of linear functions, such as limited expressiveness, easy overfitting, and poor generalization, the above implementation introduces a nonlinear function for mapping. The sigmoid function is used as the activation function mapping in the above implementation. The formula of the sigmoid function is as follows:

[0120]

[0121] The Sigmoid function can convert the input linear equation into a nonlinear output between (0, 1), which has the advantages of strong interpretability and easy convergence. Substituting the linear fourth eigenvalue (also called weight) equation into the sigmoid function, we obtain the final input weight (the correlation eigenvalue between two users) using the following calculation formula:

[0122]

[0123] It should be noted that step 102 and its corresponding implementations are the feature engineering process in data preprocessing, which includes steps such as creating new features, selecting important features, and transforming existing features. Feature engineering can enhance the model's ability to identify key information in the data, improve prediction accuracy, and improve the model's operational efficiency and generalization capabilities by reducing data complexity.

[0124] As an optional implementation, step 103 includes:

[0125] An undirected weighted graph is established based on the multiple users and the associated feature values, wherein a vertex of the undirected weighted graph represents the user, and a weight of an edge between two vertices of the undirected weighted graph represents the associated feature value; and the undirected weighted graph represents the social network.

[0126] In the above optional implementation, the user's social network is constructed based on the user's factual association (related to the user's basic information and location data) and the social circle data (the aforementioned communication data). Based on the above data attributes, the above optional implementation chooses to construct the user's social network in the form of an undirected weighted graph. The structure of the undirected weighted graph is as follows: Figure 2 As shown. Among them, Figure 2 Each node in the graph represents a user, and horizontal lines connecting nodes indicate a factual or social circle connection between the users. For example, user A and user B have a primary and secondary card relationship, user A and user C have a payment and collection relationship, and user E has no connection with users ABCD. Based on these social relationships, we can conclude that while users A and D appear to be connected only at the social circle level, they both have a strong factual connection with their mutual friend B, thus concluding that users A and D have a strong connection.

[0127] Further, Figure 2 The association relationship in the undirected weighted graph in is converted into a number (the association eigenvalue corresponding to the association relationship), and the result is as follows Figure 3 The undirected weighted graph data structure shown in Figure 1 is shown in Figure 2. Because the strength of associations between users varies, each horizontal line is weighted, with the weight representing the strength of the association. The weight (the aforementioned association eigenvalue) is calculated based on the operator's user basic information, user social circle (communication data), and user location information. During actual calculations, the above social network structure is converted into an adjacency matrix to facilitate subsequent calculations. An example of an adjacency matrix is ​​shown below:

[0128]

[0129] Each row and column of the adjacency matrix represents the strength of a user's association with other users, and each number in the matrix represents the weight (eigenvalue) of the association between users. For example, if the matrix A(2,1)=1, the association weight between user A and user B is 1. Due to the properties of undirected weighted graphs, the adjacency matrix is ​​symmetric about its diagonal.

[0130] As an optional implementation, step 104 includes:

[0131] In the social network, multiple paths between the first user to be identified and the second user to be identified are obtained; the paths here include direct paths between the first user to be identified and the second user to be identified, and also include paths passing through other users; wherein, Figure 3 For example, the paths between user A and user B include: AB path, ACB path, and ADB path.

[0132] As an example, the above steps can obtain multiple paths by traversing the associated paths between users. Among them, the iterative depth-first search algorithm (Iterative Depth-First Search, Iterative DFS) can be used to traverse / search the associated paths to obtain all paths between two users (the first user to be identified and the second user to be identified). This algorithm is a search algorithm based on the idea of ​​recursion. It is optimized on the basis of traditional DFS, avoids the problem of stack overflow caused by recursive calls, and can be used to process large-scale data sets. Among them, the logic of the iterative depth optimization search algorithm is as follows: first, create an empty stack to store the nodes to be visited; second, select the starting node and push it into the stack; third, pop the top node of the stack, traverse and visit all adjacent nodes, push the adjacent nodes into the stack and mark them as visited, and repeat this step until the stack is empty.

[0133] Calculate the association strength corresponding to each path. For a path, its association strength is the sum of the association feature values ​​of all two adjacent users on the path. For example, taking a path ABCD as an example, the association strength corresponding to this path is: the sum of the association feature values ​​of A and B, the association feature values ​​of B and C, and the association feature values ​​of C and D. Therefore, the association strength corresponding to the mth path between users AB can be expressed as follows:

[0134]

[0135] Where k represents the length of the path (if a path includes four nodes ABCD, the length of the path is 3), w i represents the association eigenvalue corresponding to the i-th segment of the path. Based on this formula, we can see that the more paths there are between two users, the higher the association eigenvalue on each path, indicating that the more common friends the users have and the closer their relationship, the stronger the association strength between the two users.

[0136] Based on the path length of each path, the weight of the association strength corresponding to each path is determined. Since the longer the (association) path between two users (the more other users it passes through), the weaker the association between the two users, this step introduces a negative penalty for the length of the path. That is, this step determines the weight of the path based on its length, for example, the weight is the inverse of the square of the path length.

[0137] According to the weights, the association strengths corresponding to the paths are weighted and summed to determine the comprehensive association strength. This step can be expressed as:

[0138]

[0139] Among them, nm represents the length of the mth path between AB, M represents the total number of paths between users AB, k represents the length of the mth path, w i Represents the associated eigenvalue corresponding to the i-th segment in the m-th path.

[0140] It should be noted that, since longer paths indicate weaker associations between two users, to reduce computational overhead, the optional implementation described above can calculate association strengths not for all paths between two users, but only for paths that meet certain conditions. Therefore, the "calculating the association strength corresponding to each path" in the optional implementation described above can specifically be: calculating the association strength corresponding to each path whose length is less than a preset value (e.g., 4). This reduces the recursive depth when traversing association paths between users.

[0141] The embodiment of the present application proposes a method for identifying a user association relationship model by constructing an operator user social network. Specifically, it is based on a graph algorithm and utilizes multiple data dimensions such as operator user basic information, call behavior, and location information to construct a user social network, and deeply mines and analyzes user social circle data to evaluate the strength of association between users. Among them, the key to this solution is to construct a user social network based on operator data, calculate the potential association between users through multiple dimensions such as user basic information, social circle, and location information, and mine common friends between users based on the user social network. Finally, the strength of association between users is judged from two levels: direct association between users and common friends.

[0142] Among them, firstly, the above-mentioned user association strength identification method of the present application identifies user association by constructing an operator user social network from the perspective of graph calculation. Compared with the current judgment of user association relationships through experience and rules, it has higher recognition accuracy; secondly, the concept of common friends is introduced on the basis of direct association between users, and the strength of the association relationship between users is comprehensively judged by the number of common friends between users and the association and density of common friends. Compared with only using dimensional data such as user social circles and location information to judge the direct association between users, it has higher judgment accuracy; thirdly, on the basis of the two dimensions of user social circles and location information, the present application introduces user basic information (ID card information, main and secondary card backbone information, collection and payment information, family number information) to judge the factual association between users, further improving the accuracy of association strength identification; fourthly, the present application introduces the concept of location trajectory overlap on the basis of judging whether users have a geographical location dimension by judging the user's permanent workplace and residence. By judging the overlap of the Nanhu trajectory position at a specific time, the accuracy of identifying the correlation between location behaviors between users can be improved.

[0143] In addition, with the continuous upgrading of illegal technologies in telecommunications networks, the behavior of criminals logging in under false names and stealing credit cards has been rampant. Traditional means are based on simple rule processing based on the operator's social circle and location data, which can only identify direct connections between users, and the granularity and accuracy of identifying related users are not high. The above embodiment of the present application introduces graph computing technology to integrate multi-dimensional data such as user basic information, social circle, and location behavior to build a user social network. Based on the user social network, the interaction behavior between users is deeply mined to discover the user's social circle. By comparing the degree of overlap between the social circles of users, the accuracy of identifying the strength of user associations can be further improved. This provides reliable data support for the judgment of the above-mentioned illegal behavior.

[0144] The embodiment of the present application provides a user association strength identification device, such as Figure 4 As shown, the device includes:

[0145] An acquisition module 401 is configured to acquire first information of a plurality of users, wherein the first information of each user includes at least one of basic user information, communication data, and location data;

[0146] A first determining module 402 is configured to determine a correlation feature value between every two users based on the first information;

[0147] A construction module 403 is configured to construct a social network of the plurality of users according to the associated feature values;

[0148] The second determining module 404 is configured to determine, based on the social network, a comprehensive correlation strength between a first user to be identified and a second user to be identified, wherein the first user to be identified and the second user to be identified are users among the plurality of users.

[0149] The first determining module 402 includes:

[0150] A first determining submodule, configured to determine, based on the first information, at least one characteristic value representing an association relationship between two users;

[0151] A second determining submodule, configured to determine the associated characteristic value based on the at least one characteristic value;

[0152] The at least one characteristic value includes at least one of the following:

[0153] a first feature value determined according to the user basic information, the first feature value being used to characterize a factual association degree between two users;

[0154] a second characteristic value determined according to the communication data, the second characteristic value being used to characterize the degree of association between the social circles of the two users;

[0155] A third eigenvalue is determined according to the location data, where the third eigenvalue is used to characterize a location association degree between the two users.

[0156] The basic user information includes at least one of user identity information, subscription package information and payment information;

[0157] The first determining submodule includes:

[0158] A first determining unit is configured to determine, based on the user basic information, if two users meet a first condition, that the first characteristic values ​​corresponding to the two users are a first value, wherein the first condition includes at least one of the following:

[0159] The user identity information bound to the two users is the same;

[0160] The subscription package indicates that the two users are family accounts;

[0161] The subscription package indicates the relationship between the two users as primary and secondary cards;

[0162] The payment information indicates that there is an agency collection and payment relationship between the two users.

[0163] The communication data includes the total number of calls, total call duration and total number of text messages between the user and different users;

[0164] The first determining submodule includes:

[0165] a processing unit, configured to normalize the total number of calls, the total duration of calls, and the total number of text messages in the communication data;

[0166] The first acquisition unit is configured to perform weighted summation on the normalized total number of calls, the total call duration, and the total number of text messages to obtain the second eigenvalue.

[0167] The first determining submodule includes:

[0168] a second determining unit, configured to determine consistency between the permanent locations of the two users based on the location data;

[0169] a second acquiring unit, configured to acquire a permanent location characteristic value corresponding to the consistency of the permanent location;

[0170] a third determining unit, configured to determine a characteristic value of a degree of overlap of the position trajectories of the two users based on the position data;

[0171] The third acquiring unit is configured to perform weighted summation on the permanent location feature value and the position trajectory coincidence feature value to obtain the third feature value.

[0172] The second determining submodule includes:

[0173] a fourth acquiring unit, configured to perform a weighted summation on at least two of the first eigenvalue, the second eigenvalue, and the third eigenvalue to obtain a fourth eigenvalue;

[0174] The fourth determining unit is configured to map the fourth eigenvalue using a nonlinear function to determine the associated eigenvalue.

[0175] The construction module 403 includes:

[0176] A submodule is established, configured to establish an undirected weighted graph based on the plurality of users and the associated feature values, wherein a vertex of the undirected weighted graph represents the user, and the weight of an edge between two vertices of the undirected weighted graph represents the associated feature value; and the undirected weighted graph represents the social network.

[0177] The second determining module 404 includes:

[0178] an acquisition submodule, configured to acquire, in the social network, a plurality of paths between the first user to be identified and the second user to be identified;

[0179] The calculation submodule is used to calculate the correlation strength corresponding to each path;

[0180] A third determining submodule is configured to determine the weight of the association strength corresponding to each path according to the path length of each path;

[0181] The fourth determining submodule is configured to perform weighted summation on the association strengths corresponding to the respective paths according to the weights to determine the comprehensive association strength.

[0182] It should be noted here that the above-mentioned user association strength identification device provided in the embodiment of the present application can implement all the method steps implemented in the above-mentioned user association strength identification method embodiment, and can achieve the same technical effect. The parts and beneficial effects of this embodiment that are the same as those in the method embodiment will not be described in detail here.

[0183] An embodiment of the present application also provides a user association strength identification device, including a transceiver 510, a processor 500, a memory 520, and a program stored on the memory 520 and executable on the processor 500; wherein, when the processor 500 executes the program, the user association strength identification device as described above is implemented.

[0184] The transceiver 510 is configured to receive and send data under the control of the processor 500 .

[0185] Among them, Figure 5 In the embodiment of the present invention, the bus architecture can include any number of interconnected buses and bridges, specifically linking various circuits such as one or more processors represented by processor 500 and memory represented by memory 520. The bus architecture can also link various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore not further described herein. The bus interface provides an interface. The transceiver 510 can be multiple components, namely, a transmitter and a receiver, providing a unit for communicating with various other devices over a transmission medium.

[0186] The processor 500 is responsible for managing the bus architecture and general processing, and the memory 520 can store data used by the processor 500 when performing operations.

[0187] The present application also provides a readable storage medium having a program stored thereon. When executed by a processor, the program implements the above-described method for identifying user association strength and achieves the same technical effect. To avoid repetition, the above description is omitted. The readable storage medium may be, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0188] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, and of course, by hardware, but in many cases the former is a better embodiment. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, disk, CD), and includes a number of instructions for executing the methods described in each embodiment of the present application.

[0189] Therefore, an embodiment of the present application also provides a computer program product, including computer instructions, which, when executed by a processor, implement the user association strength identification method as described above and can achieve the same technical effect. To avoid repetition, they will not be described here.

[0190] In embodiments of the present application, modules can be implemented in software so that they can be executed by various types of processors. For example, an identified executable code module can include one or more physical or logical blocks of computer instructions, for example, which can be constructed as objects, processes, or functions. Nevertheless, the executable code of the identified module does not need to be physically located together, but can include different instructions stored in different locations, which, when logically combined together, constitute the module and achieve the specified purpose of the module.

[0191] In fact, executable code module can be a single instruction or many instructions, and can even be distributed on a plurality of different code segments, distributed in the middle of different programs, and distributed across a plurality of memory devices.Similarly, operating data can be identified in the module, and can be implemented and organized in the data structure of any appropriate type according to any appropriate form.Described operating data can be collected as a single data set, or can be distributed in different locations (including on different storage devices), and can only be present on a system or network as an electronic signal at least in part.

[0192] When a module can be implemented using software, given the current state of hardware technology, those skilled in the art can build corresponding hardware circuits to implement the corresponding functions of the module, regardless of cost. The hardware circuits may include conventional very large scale integration (VLSI) circuits or gate arrays, as well as existing semiconductors such as logic chips and transistors, or other discrete components. Modules may also be implemented using programmable hardware devices, such as field programmable gate arrays, programmable array logic, or programmable logic devices.

[0193] The above exemplary embodiments are described with reference to the accompanying drawings. Many different forms and embodiments are possible without departing from the spirit and teachings of this application. Therefore, this application should not be construed as limited to the exemplary embodiments set forth herein. Rather, these exemplary embodiments are provided so that this application will be complete and impartial and will convey the scope of this application to those skilled in the art. In the drawings, component sizes and relative sizes may be exaggerated for clarity. The terminology used herein is for purposes of describing specific exemplary embodiments only and is not intended to be limiting. As used herein, unless the context clearly indicates otherwise, the singular forms "a," "an," and "the" are intended to encompass such plural forms. It will be further understood that the terms "comprising" and / or "including," when used in this specification, indicate the presence of the stated features, integers, steps, operations, components, and / or elements, but do not preclude the presence or addition of one or more other features, integers, steps, operations, components, elements, and / or groups thereof. Unless otherwise indicated, when stated, a range of values ​​includes the upper and lower limits of that range and any subranges therebetween.

[0194] The above is a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles described in the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for identifying user association strength, characterized in that: include: Acquire first information of a plurality of users, where the first information of each user includes at least one of user basic information, communication data, and location data; Determining, based on the first information, a correlation feature value between every two users; constructing a social network of a plurality of the users according to the associated feature values; Based on the social network, a comprehensive correlation strength between a first user to be identified and a second user to be identified is determined, wherein the first user to be identified and the second user to be identified are users among the plurality of users.

2. The method according to claim 1, characterized in that Determining, based on the first information, an association feature value between every two users, including: Determining, based on the first information, at least one characteristic value representing an association relationship between two users; determining the associated feature value based on the at least one feature value; The at least one characteristic value includes at least one of the following: a first feature value determined according to the user basic information, the first feature value being used to characterize a factual association degree between two users; a second characteristic value determined according to the communication data, the second characteristic value being used to characterize the degree of association between the social circles of the two users; A third eigenvalue is determined according to the location data, where the third eigenvalue is used to characterize a location association degree between the two users.

3. The method according to claim 2, characterized in that The basic user information includes at least one of user identity information, subscription package information and payment information; Determining, based on the first information, at least one characteristic value representing an association relationship between two users includes: According to the user basic information, if two users meet a first condition, determining that the first characteristic values ​​corresponding to the two users are a first value, wherein the first condition includes at least one of the following: The user identity information bound to the two users is the same; The subscription package indicates that the two users are family accounts; The subscription package indicates the relationship between the two users as primary and secondary cards; The payment information indicates that there is an agency collection and payment relationship between the two users.

4. The method according to claim 2, characterized in that The communication data includes the total number of calls, total call duration, and total number of text messages between the user and different users; Determining, based on the first information, at least one characteristic value representing an association relationship between two users includes: Normalizing the total number of calls, the total duration of calls, and the total number of text messages in the communication data; The normalized total number of calls, the total call duration, and the total number of text messages are weightedly summed to obtain the second eigenvalue.

5. The method according to claim 2, characterized in that Determining, based on the first information, at least one characteristic value representing an association relationship between two users includes: Determining consistency between the permanent locations of the two users based on the location data; obtaining a permanent location characteristic value corresponding to the consistency of the permanent location; Determining, based on the location data, a characteristic value of a degree of overlap of the location trajectories of the two users; A weighted sum is performed on the permanent location eigenvalue and the position trajectory coincidence eigenvalue to obtain the third eigenvalue.

6. The method according to claim 2, characterized in that Determining the associated characteristic value according to the at least one characteristic value includes: performing a weighted summation on at least two of the first eigenvalue, the second eigenvalue, and the third eigenvalue to obtain a fourth eigenvalue; The fourth eigenvalue is mapped using a nonlinear function to determine the associated eigenvalue.

7. The method according to claim 1, characterized in that Constructing a social network of multiple users according to the associated feature values, including: An undirected weighted graph is established based on the multiple users and the associated feature values, wherein a vertex of the undirected weighted graph represents the user, and a weight of an edge between two vertices of the undirected weighted graph represents the associated feature value; and the undirected weighted graph represents the social network.

8. The method according to claim 1, characterized in that Determining, based on the social network, a comprehensive correlation strength between the first user to be identified and the second user to be identified, including: In the social network, obtaining multiple paths between the first user to be identified and the second user to be identified; Calculate the correlation strength corresponding to each path; Determining the weight of the association strength corresponding to each path according to the path length of each path; The association strengths corresponding to the various paths are weighted and summed according to the weights to determine the comprehensive association strength.

9. A user association strength identification device, characterized in that: include: An acquisition module, configured to acquire first information of a plurality of users, wherein the first information of each user includes at least one of basic user information, communication data, and location data; A first determining module, configured to determine an association feature value between every two users based on the first information; A construction module, configured to construct a social network of a plurality of the users according to the associated feature values; The second determining module is configured to determine, based on the social network, a comprehensive correlation strength between a first user to be identified and a second user to be identified, wherein the first user to be identified and the second user to be identified are users among the plurality of users.

10. A user association strength recognition device, comprising a transceiver, a processor, a memory, and a program stored in the memory and executable on the processor; characterized in that: The transceiver is used to receive and send data under the control of the processor, and the processor implements the user association strength identification method according to any one of claims 1 to 8 when executing the program.

11. A readable storage medium having a program or instruction stored thereon, characterized in that: When the program or instruction is executed by a processor, the user association strength identification method according to any one of claims 1 to 8 is implemented.

12. A computer program product, characterized in that The method comprises computer instructions, which, when executed by a processor, implements the user association strength identification method according to any one of claims 1 to 8.

Images