A social relationship evaluation method and device

Abstract

The present invention provides a social relationship evaluation method and device, the method comprising: obtaining social feature information of a user to be evaluated, and obtaining social feature information of at least one neighboring user whose distance from the user to be evaluated is less than the execution distance in a preset social network; inputting the social feature information of the user to be evaluated and the social feature information of the neighboring user into a pre-constructed social relationship evaluation model; using the social relationship evaluation model to evaluate the social risk of the user to be evaluated according to the social feature information of the user to be evaluated and the social feature information of the neighboring user, and outputting the evaluation result of the social risk level of the user to be evaluated. Based on the scheme provided by the present invention, the social risk of the user to be evaluated can be accurately and quickly evaluated, providing a strong basis for the credit of the user to be evaluated in some security information fields.

Description

Technical Field

[0001] The present invention relates to the field of Internet technology, and in particular to a social relationship evaluation method and device. Background Art

[0002] Risk control at the level of social network relationships is an important source of credit and basis in the field of micro-loans and anti-fraud. Relationship network modeling is usually based on the meta-principle that birds of a feather flock together. Representative technologies under this idea include community mining, label propagation and other technologies.

[0003] However, social networks are usually large in size and it is difficult to distinguish good and bad people, and it is difficult to obtain reasonable labels. Therefore, clustering based on label propagation algorithms is prone to misjudgment and cannot be directly used as an independent evaluation model. In addition, relying solely on the characteristics of the node relationship level cannot deeply explore the user's internal information and cannot comprehensively judge information from multiple dimensions. Summary of the invention

[0004] The present invention provides a social relationship evaluation method and device to overcome the above problems or at least partially solve the above problems.

[0005] According to one aspect of the present invention, a social relationship evaluation method is provided, comprising:

[0006] Acquire social feature information of the user to be evaluated, and acquire social feature information of at least one neighboring user whose distance to the user to be evaluated is less than the execution distance in a preset social network;

[0007] Inputting the social feature information of the user to be evaluated and the social feature information of the neighboring users into a pre-built social relationship evaluation model;

[0008] The social relationship evaluation model is used to evaluate the social risk of the user to be evaluated according to the social feature information of the user to be evaluated and the social feature information of the neighboring users, and an evaluation result of the social risk level of the user to be evaluated is output.

[0009] Optionally, before acquiring social feature information of the user to be evaluated and acquiring social feature information of at least one neighboring user whose distance to the user to be evaluated is less than the execution distance in a preset social network, the method further includes:

[0010] Collecting social data of multiple users and building a user social database; wherein the social data includes: social data generated based on telecommunication communication, network communication and / or offline social interaction;

[0011] A social network is created, each user is mapped to each node of the social network based on the social data of each user, and a social connection is established between users having a social relationship.

[0012] Optionally, the creating of the social network, after mapping each user to each node of the social network based on the social data of each user and establishing social connections between users having social relationships, further includes:

[0013] A user portrait is created for each user based on the social data of each user in the user database, and a user tag indicating a social risk level is added to at least a specified proportion of users in the user database.

[0014] Optionally, before inputting the social feature information of the user to be evaluated and the social feature information of the neighboring users into a pre-built social relationship evaluation model, the method further includes:

[0015] Construct a social relationship evaluation model;

[0016] Semi-supervised learning is performed based on each user in the user database and the social data corresponding to each user to train the social relationship evaluation model.

[0017] Optionally, performing semi-supervised learning based on each user in the user database and social data corresponding to each user to train the social relationship evaluation model includes:

[0018] Sampling is performed based on the user social database, and all users included in the user social database are divided into multiple sampling groups according to preset conditions;

[0019] For any group of sampled groups, obtain social feature information of each user in the sampled group, and generate graph relationship data of the sampled group based on the social feature information of each user in the sampled group;

[0020] Semi-supervised learning is performed on the graph relationship data of each group of sampled groups according to graph convolution to train the social relationship evaluation model.

[0021] Optionally, the preset condition includes at least one of the following:

[0022] Each user in each sampling group has a social connection with at least one other user in the sampling group to which it belongs;

[0023] There is overlap between different sampling groups;

[0024] The users in each sampling group who have the user label identifying the social risk level are within a specified proportion range of all users in the sampling group;

[0025] The number of users included in each sampling group is within a preset number;

[0026] Any user in the user social database belongs to at least one sampling group.

[0027] Optionally, for any group of sampled groups, obtaining social feature information of each user in the sampled group, and generating graph relationship data of the sampled group based on the social feature information of each user in the sampled group includes:

[0028] For any group of sampling groups, obtaining the social feature information of each user in the sampling group from the social data of each user in the sampling group according to the first weight rule;

[0029] The graph relationship data of the sample group is generated based on the social feature information of each user.

[0030] Optionally, the acquiring of social feature information of the user to be evaluated, and acquiring social feature information of at least one neighboring user whose distance from the user to be evaluated is less than the execution distance in a preset social network, includes:

[0031] According to the second weight rule, social feature information of the user to be evaluated and the neighboring user is obtained from the social data of the user to be evaluated and at least one neighboring user whose distance to the user to be evaluated is less than the execution distance.

[0032] According to another aspect of the present invention, there is also provided a social relationship evaluation device, comprising:

[0033] An acquisition module configured to acquire social feature information of the user to be evaluated, and acquire social feature information of at least one neighboring user whose distance to the user to be evaluated is less than an execution distance in a preset social network;

[0034] An input module, configured to input the social feature information of the user to be evaluated and the social feature information of the neighboring users into a pre-built social relationship evaluation model;

[0035] The evaluation module is configured to evaluate the social risk of the user to be evaluated according to the social feature information of the user to be evaluated and the social feature information of the neighboring users by using the social relationship evaluation model, and output an evaluation result of the social risk level of the user to be evaluated.

[0036] Optionally, the device further comprises:

[0037] A creation module is configured to collect social data of multiple users and build a user social database; wherein the social data includes: social data generated based on telecommunication communication, network communication and / or offline social interaction;

[0038] A social network is created, each user is mapped to each node of the social network based on the social data of each user, and a social connection is established between users having a social relationship.

[0039] Optionally, the creation module further configures:

[0040] A user portrait is created for each user based on the social data of each user in the user database, and a user tag indicating a social risk level is added to at least a specified proportion of users in the user database.

[0041] Optionally, the device further comprises:

[0042] A building module configured to build a social relationship evaluation model;

[0043] Semi-supervised learning is performed based on each user in the user database and the social data corresponding to each user to train the social relationship evaluation model.

[0044] Optionally, the building module is further configured as:

[0045] Sampling is performed based on the user social database, and all users included in the user social database are divided into multiple sampling groups according to preset conditions;

[0046] For any group of sampled groups, obtain social feature information of each user in the sampled group, and generate graph relationship data of the sampled group based on the social feature information of each user in the sampled group;

[0047] Semi-supervised learning is performed on the graph relationship data of each group of sampled groups according to graph convolution to train the social relationship evaluation model.

[0048] Optionally, the preset condition includes at least one of the following:

[0049] Each user in each sampling group has a social connection with at least one other user in the sampling group to which it belongs;

[0050] There is overlap between different sampling groups;

[0051] The users in each sampling group who have the user label identifying the social risk level are within a specified proportion range of all users in the sampling group;

[0052] The number of users included in each sampling group is within a preset number;

[0053] Any user in the user social database belongs to at least one sampling group.

[0054] Optionally, the building module is further configured as:

[0055] For any group of sampling groups, obtaining the social feature information of each user in the sampling group from the social data of each user in the sampling group according to the first weight rule;

[0056] The graph relationship data of the sample group is generated based on the social feature information of each user.

[0057] Optionally, the acquisition module is further configured to:

[0058] According to the second weight rule, social feature information of the user to be evaluated and the neighboring user is obtained from the social data of the user to be evaluated and at least one neighboring user whose distance to the user to be evaluated is less than the execution distance.

[0059] According to another aspect of the present invention, a computer storage medium is provided, wherein the computer storage medium stores a computer program code, and when the computer program code is executed on a computing device, the computing device is caused to execute any one of the above-mentioned social relationship evaluation methods.

[0060] According to another aspect of the present invention, there is further provided a computing device, comprising:

[0061] processor;

[0062] a memory storing computer program code;

[0063] When the computer program code is executed by the processor, the computing device is caused to execute any one of the above-mentioned social relationship evaluation methods.

[0064] The present invention provides a more efficient social relationship evaluation method and device, which obtains the social feature information of the user to be evaluated and the social feature information of at least one neighboring user, and then inputs the social feature information obtained into a pre-built social relationship evaluation model, evaluates the social risk of the user to be evaluated, and then outputs the evaluation result of the social risk level of the user to be evaluated. Based on the method provided by the present invention, the existing social relationship evaluation model is used to analyze the social feature information of the user to be evaluated and its neighboring users, which can combine the scarce label information, the massive user relationship network data, and the massive user portrait information, and then accurately and quickly evaluate the social risk of the user to be evaluated, providing a strong basis for the credit of the user to be evaluated in some security information fields.

[0065] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below.

[0066] Based on the following detailed description of specific embodiments of the present invention in conjunction with the accompanying drawings, those skilled in the art will become more aware of the above and other objects, advantages and features of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0068] Figure 1 is a flowchart of a social relationship evaluation method according to an embodiment of the present invention;

[0069] Figure 2 is a schematic diagram of a graph relationship corresponding to a sampling group according to an embodiment of the present invention;

[0070] Figure 3 is a schematic diagram of the structure of a social relationship evaluation device according to an embodiment of the present invention;

[0071] Figure 4 Schematic diagram of the structure of a social relationship evaluation device according to a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0072] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0073] Figure 1 A schematic diagram of a social relationship evaluation method according to an embodiment of the present invention is shown. Figure 1 It can be seen that the social relationship evaluation method provided by the embodiment of the present invention may include:

[0074] Step S102, obtaining social feature information of the user to be evaluated, and obtaining social feature information of at least one neighboring user whose distance to the user to be evaluated is less than the execution distance in a preset social network;

[0075] Step S104, inputting the social characteristic information of the user to be evaluated and the social characteristic information of the neighboring users into a pre-built social relationship evaluation model;

[0076] Step S106, using the social relationship evaluation model to evaluate the social risk of the user to be evaluated according to the social characteristic information of the user to be evaluated and the social characteristic information of the neighboring users, and outputting an evaluation result of the social risk level of the user to be evaluated.

[0077] The embodiment of the present invention provides a more efficient social relationship evaluation method, which obtains the social feature information of the user to be evaluated and the social feature information of at least one neighboring user, and then inputs the social feature information obtained into a pre-built social relationship evaluation model, evaluates the social risk of the user to be evaluated, and then outputs the evaluation result of the social risk level of the user to be evaluated. Based on the method provided by the embodiment of the present invention, the existing social relationship evaluation model is used to analyze the social feature information of the user to be evaluated and its neighboring users, and then the social risk of the user to be evaluated is accurately and quickly evaluated, providing a strong basis for the credit of the user to be evaluated in some security information fields.

[0078] As mentioned in the above step S102, at least one neighboring user of the user to be evaluated is obtained in a preset social network. Optionally, before the above step S102, the following steps may also be included:

[0079] S1-1, collecting social data of multiple users and building a user social database; wherein the social data includes: social data generated based on telecommunication communication, network communication and / or offline social interaction;

[0080] S1-2, creating a social network, mapping each user to each node of the social network based on the social data of each user, and establishing social connections between users with social relationships.

[0081] The social relationship between two users can often be roughly inferred based on the frequency, method and time of their contact. For example, two users who interact frequently can be relatives, friends, couples, etc. In practical applications, users' social data is massive, and the social methods are multi-dimensional. Therefore, the social data of the embodiment of the present invention can be the collection of social data of multiple (massive) users, such as telecommunications, network communications and / or offline social data generated to build a social database. Among them, telecommunications communications are social methods through traditional telephones and text messages, network communications can include instant messaging software, such as WeChat, QQ, email social methods, or social methods through other applications; offline social data generated by social data can be social data generated by meeting, participating in social activities together, etc. In addition to users and corresponding social data, the user social database can also add information such as the user's social attributes.

[0082] After collecting the social data of a large number of users, a social network can be created, and then each user can be mapped to the social network as a node in the social network. Among them, the dimensions of the social network can be divided in a social way, or divided according to other methods, which is not limited by the present invention. After all users are mapped to the social network, a social connection can be established between users with social relationships based on the social data of each user. For any user, the user with a social relationship with the user can be a user in the local address book of the user, or a user with a friend relationship with the user in one or more network communication applications, or a user with email contact. Establishing a social connection can be connecting two users in a social network with a connecting line or other methods, and the social connection for determining social relationships in different ways can be changed. In addition to the above introduction, in a social network, the distance between two users with a social connection can be calculated based on the social data of both parties. The more social data between two users, the closer the relationship between the two users. At this time, the distance between the two users in the social network can also be shorter.

[0083] Optionally, after the above step S1-2, the following steps may also be included:

[0084] S1-3, creating a user portrait for each user based on the social data of each user in the user database, and adding user tags that identify the social risk level to at least a specified proportion of users in the user database.

[0085] User portrait is a labeled user model abstracted from the user's social attributes, living habits, interests and hobbies. The process of establishing user portrait is to add corresponding labels to users, which is called labeling in the field of data mining. Labels are highly refined feature identifiers obtained by analyzing user information.

[0086] In the embodiment of the present invention, based on the collected user social data, user labels that identify the social risk level, such as risk-free users, risky users, and suspected dangerous users, can be added to the user based on social data of various dimensions, such as the user's communication information, phone number tag information, and other social related information. When adding, the main thing is to first add corresponding user labels to users whose social risk level can be directly determined. In addition, since the user's social data can be increased based on different time periods, the user social data in the user social database in the embodiment of the invention can be continuously updated, and the corresponding user portraits and user labels can also be continuously updated.

[0087] In the above step S104, it is mentioned that after obtaining the social feature information of the user to be evaluated and the social feature information of the neighboring users, the above information will be input into the pre-built social relationship evaluation model. Therefore, before this, it can also include: building a social relationship evaluation network model; performing semi-supervised learning based on each user in the user database and the social data corresponding to each user, and training the social network evaluation model.

[0088] In an optional embodiment of the present invention, the training of the social relationship evaluation model can be performed in the following manner:

[0089] S2-1, performing sampling based on a user social database, dividing all users included in the user social database into multiple sampling groups according to preset conditions;

[0090] S2-2, for any group of sampling groups, obtaining social feature information of each user in the sampling group, and generating graph relationship data of the sampling group based on the social feature information of each user in the sampling group;

[0091] S2-3, based on graph convolution, semi-supervised learning is performed on the graph relationship data of each group of sampling groups to train the social relationship evaluation model.

[0092] In the embodiment of the present invention, users included in the social network are first divided into groups, and a large sampling group is divided into multiple small sampling groups. The preset conditions for the division of each sampling group may include at least one of the following:

[0093] 1. Each user in each sampling group has a social connection with at least one other user in the sampling group to which it belongs;

[0094] 2. There are overlaps between different sampling groups; that is, any sampling group may have one or more common user nodes with one or more other sampling groups;

[0095] 3. The users with the user labels identifying the social risk level in each sampling group are the specified proportion range of all users in each sampling group; since the user labels identifying the social risk level may be a minority, therefore, when dividing the sampling groups, it is ensured that there are a certain number of users carrying the user labels in each sampling group, so that the model trained based on the above data is more accurate when evaluating the social relationships of unknown users; in a sampling group, the specified proportion range of known users carrying the above user labels can be set according to the actual number of actual known users and the total number of users, and the present invention does not limit it;

[0096] 4. The number of users included in each sampling group is within a preset number; wherein the preset number can be set based on different scenarios;

[0097] 5. Any user in the user social database belongs to at least one sampling group.

[0098] In an embodiment of the present invention, it is necessary to first construct a social relationship evaluation model, and then use the social data corresponding to the divided sampling group to train the social relationship evaluation model. When constructing the social relationship evaluation model, the model can be constructed based on a neural network system. For model training, there are four main methods commonly used: supervised, unsupervised, semi-supervised and reinforcement learning. The semi-supervised learning used in the embodiment of the present invention is a learning method that combines supervised learning with unsupervised learning. Semi-supervised learning uses a large amount of unlabeled data, as well as labeled data at the same time, to perform pattern recognition.

[0099] As introduced above, when the extraction groups are divided, each extraction group includes users with user labels that identify social risk levels within a specified proportion range. That is, the social data of users with user labels that identify social risk levels are used as labeled data, and the social data of other users without similar user labels are used as unlabeled data. The above two types of data are used to train the social relationship evaluation model, thereby making the evaluation accuracy and efficiency of the social relationship evaluation model higher.

[0100] It is mentioned in step S2-2 that for any group of sampling groups, the graph relationship data generated by the social feature information of each user in the sampling group is then used as the training data of the social relationship evaluation model. Among them, the social feature information can be generated based on the social data of the user. Optionally, when determining the training data of the social relationship evaluation model, the social feature information of each user in the sampling group can be obtained according to the first weight rule from the social data of each user in the sampling group; the graph relationship data of the sampling group is generated based on the social feature information of each user. The first weight rule in this embodiment may include the weight ratio of the social data of each dimension in the social data of each user, and can be set in combination with the number of social objects of the user, the social attributes of each social object, etc.

[0101] The graph relationship data of the sample group generated based on the social feature information of each user can be as follows Figure 2 As shown, Figure 2 Each user is used as a connection node, and social data of different dimensions use graph relational data with different connection methods. Figure 2In the graph relationship data shown, 1, 2, 3, 4, 5, and 6 represent 6 users in the sample group, respectively. Offline social networking, instant messaging software social networking, and telecommunications call social networking are represented by solid line a, dotted line b, and solid line dotted line interval line c, respectively. In addition, user labels of different social risk levels can also be displayed in black or white in the graph. If a user's social risk level user label is dangerous, it is represented in black. If a user's social risk level user label is safe, it is represented in white. Of course, potentially dangerous users can also be represented in gray or in other ways. After obtaining the graph relationship data of a sample group in the above manner, it can be input into a pre-built social relationship evaluation model for training.

[0102] The social feature information of the user to be evaluated and the social feature information of the neighboring users obtained in the above step S102 can be obtained in the social data of the user to be evaluated and the neighboring users according to the second weight rule. The second weight rule can be the same as the first weight rule, or it can be set according to different application scenarios. The second weight rule can also be adjusted accordingly according to the social attributes of the user, and the present invention does not limit it. For example, for business use, the weight of email-related social data may be larger, and for sales-type users, the weight of phone-related social data may be larger, and so on, which will not be repeated here. The above-mentioned acquired feature information is input into the trained social relationship evaluation model. The social relationship evaluation model can analyze and judge the feature information of the user to be evaluated and the neighboring users, and output the social risk probability of the user to be evaluated, for example, 90% is a good person, and then add a user label to the user to identify the social risk level of the user to be evaluated based on the input social risk probability, so as to provide powerful data for the subsequent risk control review of the user to be evaluated.

[0103] Based on the same inventive concept, the embodiment of the present invention also provides a social relationship evaluation device, such as Figure 3 As shown, the social relationship evaluation device provided according to an embodiment of the present invention may include:

[0104] The acquisition module 310 is configured to acquire social feature information of the user to be evaluated, and acquire social feature information of at least one neighboring user whose distance to the user to be evaluated is less than the execution distance in a preset social network;

[0105] An input module 320, configured to input the social characteristic information of the user to be evaluated and the social characteristic information of the neighboring users into a pre-built social relationship evaluation model;

[0106] The evaluation module 330 is configured to evaluate the social risk of the user to be evaluated based on the social feature information of the user to be evaluated and the social feature information of the neighboring users using the social relationship evaluation model, and output an evaluation result of the social risk level of the user to be evaluated.

[0107] In an optional embodiment of the present invention, Figure 4 As shown, the above device may also include:

[0108] A creation module 340 is configured to collect social data of multiple users and construct a user social database; wherein the social data includes: social data generated based on telecommunication communication, network communication and / or offline social interaction;

[0109] Create a social network, map each user to a node of the social network based on the social data of each user, and establish social connections between users with social relationships.

[0110] In an optional embodiment of the present invention, the creation module 340 may also be configured as follows:

[0111] A user profile is created for each user based on the social data of each user in the user database, and a user label indicating a social risk level is added to at least a specified proportion of users in the user database.

[0112] In an optional embodiment of the present invention, Figure 4 As shown, the above device may also include:

[0113] A construction module 350 is configured to construct a social relationship evaluation model;

[0114] Semi-supervised learning is performed based on each user in the user database and the social data corresponding to each user to train a social relationship evaluation model.

[0115] In an optional embodiment of the present invention, the construction module 350 may also be configured as:

[0116] Sampling is performed based on a user social database, and all users included in the user social database are divided into multiple sampling groups according to preset conditions;

[0117] For any group of sampling groups, obtain the social feature information of each user in the sampling group, and generate graph relationship data of the sampling group based on the social feature information of each user in the sampling group;

[0118] Based on graph convolution, semi-supervised learning is performed on the graph relationship data of each group of sampling groups to train the social relationship evaluation model.

[0119] In an optional embodiment of the present invention, the above-mentioned preset condition includes at least one of the following:

[0120] Each user in each sampling group has a social connection with at least one other user in the sampling group to which it belongs;

[0121] There is overlap between different sampling groups;

[0122] The users with user labels identifying social risk levels in each sampling group are within a specified proportion range of all users in the sampling group;

[0123] The number of users included in each sampling group is within a preset number;

[0124] Any user in the user social database belongs to at least one sampling group.

[0125] In an optional embodiment of the present invention, the construction module 350 may also be configured as:

[0126] For any group of sampling groups, obtaining social feature information of each user from the social data of each user in the sampling group according to the first weight rule;

[0127] Generate graph relationship data of the sampled group based on the social feature information of each user.

[0128] In an optional embodiment of the present invention, the acquisition module 310 may also be configured as follows:

[0129] According to the second weight rule, social feature information of the user to be evaluated and the neighboring user is obtained in the social data of the user to be evaluated and at least one neighboring user whose distance to the user to be evaluated is less than the execution distance.

[0130] Based on the same inventive concept, an embodiment of the present invention further provides a computer storage medium, which stores computer program code. When the computer program code runs on a computing device, it causes the computing device to execute the social relationship evaluation method described in any of the above embodiments.

[0131] Based on the same inventive concept, an embodiment of the present invention further provides a computing device, including:

[0132] processor;

[0133] a memory storing computer program code;

[0134] When the computer program code is executed by a processor, it causes the computing device to execute the social relationship evaluation method described in any of the above embodiments.

[0135] The embodiment of the present invention provides a more efficient social relationship evaluation method and device, which obtains the social feature information of the user to be evaluated and the social feature information of at least one neighboring user, and then inputs the obtained social feature information into a pre-built social relationship evaluation model, evaluates the social risk of the user to be evaluated, and then outputs the evaluation result of the social risk level of the user to be evaluated. Based on the method provided by the embodiment of the present invention, the existing social relationship evaluation model is used to analyze the social feature information of the user to be evaluated and its neighboring users, and then the social risk of the user to be evaluated is accurately and quickly evaluated, providing a strong basis for the credit of the user to be evaluated in some security information fields.

[0136] Those skilled in the art can clearly understand that the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments, and for the sake of brevity, they are not further described here.

[0137] In addition, the functional units in various embodiments of the present invention may be physically independent of each other, or two or more functional units may be integrated together, or all functional units may be integrated into one processing unit. The above integrated functional units may be implemented in the form of hardware, or in the form of software or firmware.

[0138] Those skilled in the art can understand that if the integrated functional unit is implemented in the form of software and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention can essentially or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, which includes a number of instructions to enable a computing device (such as a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention when running the instructions. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk and other media that can store program codes.

[0139] Alternatively, all or part of the steps of implementing the aforementioned method embodiments may be accomplished by hardware associated with program instructions (such as a computing device such as a personal computer, a server, or a network device), and the program instructions may be stored in a computer-readable storage medium. When the program instructions are executed by a processor of a computing device, the computing device executes all or part of the steps of the methods described in the embodiments of the present invention.

[0140] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that within the spirit and principles of the present invention, the technical solutions described in the aforementioned embodiments can still be modified, or some or all of the technical features therein can be replaced by equivalents. However, these modifications or replacements do not deviate from the protection scope of the present invention.

Claims

1. A social relationship evaluation method, comprising: Acquire social feature information of the user to be evaluated, and acquire social feature information of at least one neighboring user whose distance to the user to be evaluated is less than the execution distance in a preset social network; Inputting the social feature information of the user to be evaluated and the social feature information of the neighboring users into a pre-built social relationship evaluation model; The social relationship evaluation model is used to evaluate the social risk of the user to be evaluated according to the social feature information of the user to be evaluated and the social feature information of the neighboring users, and an evaluation result of the social risk level of the user to be evaluated is output.

2. The method according to claim 1, wherein: Before acquiring the social feature information of the user to be evaluated and acquiring the social feature information of at least one neighboring user whose distance to the user to be evaluated is less than the execution distance in the preset social network, the method further includes: Collecting social data of multiple users and building a user social database; wherein the social data includes: social data generated based on telecommunication communication, network communication and / or offline social interaction; A social network is created, each user is mapped to each node of the social network based on the social data of each user, and a social connection is established between users having a social relationship.

3. The method according to claim 2, wherein: The creating of the social network, after mapping each user to each node of the social network based on the social data of each user and establishing social connections between users having social relationships, further includes: A user portrait is created for each user based on the social data of each user in the user database, and a user tag indicating a social risk level is added to at least a specified proportion of users in the user database.

4. The method according to claim 2, wherein: Before inputting the social feature information of the user to be evaluated and the social feature information of the neighboring users into a pre-built social relationship evaluation model, the method further includes: Construct a social relationship evaluation model; Semi-supervised learning is performed based on each user in the user database and the social data corresponding to each user to train the social relationship evaluation model.

5. The method according to claim 4, wherein: The performing semi-supervised learning based on each user in the user database and the social data corresponding to each user to train the social relationship evaluation model includes: Sampling is performed based on the user social database, and all users included in the user social database are divided into multiple sampling groups according to preset conditions; For any group of sampled groups, obtain social feature information of each user in the sampled group, and generate graph relationship data of the sampled group based on the social feature information of each user in the sampled group; Semi-supervised learning is performed on the graph relationship data of each group of sampled groups based on graph convolution to train the social relationship evaluation model.

6. The method according to claim 5, wherein: The preset condition includes at least one of the following: Each user in each sampling group has a social connection with at least one other user in the sampling group to which it belongs; There is overlap between different sampling groups; The users with user labels identifying social risk levels in each sampling group are within a specified proportion range of all users in the sampling group; The number of users included in each sampling group is within a preset number; Any user in the user social database belongs to at least one sampling group.

7. The method according to claim 5, wherein: For any group of sampled groups, obtaining social feature information of each user in the sampled group, and generating graph relationship data of the sampled group based on the social feature information of each user in the sampled group, includes: For any group of sampling groups, obtaining the social feature information of each user in the sampling group from the social data of each user in the sampling group according to the first weight rule; The graph relationship data of the sample group is generated based on the social feature information of each user.

8. The method according to any one of claims 2 to 7, wherein: The step of obtaining social feature information of the user to be evaluated, and obtaining social feature information of at least one neighboring user whose distance from the user to be evaluated is less than the execution distance in a preset social network, includes: According to the second weight rule, social feature information of the user to be evaluated and the neighboring user is obtained from the social data of the user to be evaluated and at least one neighboring user whose distance to the user to be evaluated is less than the execution distance.

9. A social relationship evaluation device, comprising: An acquisition module configured to acquire social feature information of the user to be evaluated, and acquire social feature information of at least one neighboring user whose distance to the user to be evaluated is less than an execution distance in a preset social network; An input module, configured to input the social feature information of the user to be evaluated and the social feature information of the neighboring users into a pre-built social relationship evaluation model; The evaluation module is configured to evaluate the social risk of the user to be evaluated according to the social feature information of the user to be evaluated and the social feature information of the neighboring users by using the social relationship evaluation model, and output an evaluation result of the social risk level of the user to be evaluated.

10. The device according to claim 9, wherein: Also includes: A creation module is configured to collect social data of multiple users and build a user social database; wherein the social data includes: social data generated based on telecommunication communication, network communication and / or offline social interaction; A social network is created, each user is mapped to each node of the social network based on the social data of each user, and a social connection is established between users having a social relationship.

11. The device according to claim 10, wherein: The creation module also configures: A user portrait is created for each user based on the social data of each user in the user database, and a user tag indicating a social risk level is added to at least a specified proportion of users in the user database.

12. The device according to claim 11, wherein: Also includes: A building module configured to build a social relationship evaluation model; Semi-supervised learning is performed based on each user in the user database and the social data corresponding to each user to train the social relationship evaluation model.

13. The device according to claim 12, wherein: The building block is further configured as: Sampling is performed based on the user social database, and all users included in the user social database are divided into multiple sampling groups according to preset conditions; For any group of sampled groups, obtain social feature information of each user in the sampled group, and generate graph relationship data of the sampled group based on the social feature information of each user in the sampled group; Semi-supervised learning is performed on the graph relationship data of each group of sampled groups based on graph convolution to train the social relationship evaluation model.

14. The device according to claim 13, wherein: The preset condition includes at least one of the following: Each user in each sampling group has a social connection with at least one other user in the sampling group to which it belongs; There is overlap between different sampling groups; The users in each sampling group who have the user label identifying the social risk level are within a specified proportion range of all users in the sampling group; The number of users included in each sampling group is within a preset number; Any user in the user social database belongs to at least one sampling group.

15. The device according to claim 13, wherein: The building block is further configured as: For any group of sampling groups, obtaining the social feature information of each user in the sampling group from the social data of each user in the sampling group according to the first weight rule; The graph relationship data of the sample group is generated based on the social feature information of each user.

16. The device according to any one of claims 10 to 15, wherein: The acquisition module is further configured as follows: According to the second weight rule, social feature information of the user to be evaluated and the neighboring user is obtained from the social data of the user to be evaluated and at least one neighboring user whose distance to the user to be evaluated is less than the execution distance.

17. A computer storage medium storing a computer program code, wherein when the computer program code is executed on a computing device, the computing device is caused to execute the social relationship evaluation method according to any one of claims 1 to 8.

18. A computing device comprising: processor; a memory storing computer program code; When the computer program code is executed by the processor, it causes the computing device to execute the social relationship evaluation method according to any one of claims 1 to 8.

Images