Key personnel relationship network analysis method, device and equipment

Abstract

The invention discloses a key person relationship network analysis method, device and equipment, and the method comprises the steps: obtaining the person contact information of different key persons based on the basic information of different key persons in a database; taking each key person and the close contact person having the contact relationship with the key person as nodes, adopting edge connection, and taking the close contact degree of the key person and the close contact person as the weight of the edge to obtain a key person relationship graph; dividing the key person relation graph into a plurality of communities, measuring the degree of close connection among nodes in the communities based on the weights of edges among the nodes in the communities to obtain the modularity of the communities, and optimizing community division; taking the personnel attribute information of the nodes of the community as samples, determining the clustering number of a sample set, and performing clustering; and after clustering is finished, performing feature analysis on the personnel attribute information in each cluster. Therefore, the problems that existing source data is relatively single, potential social relations are difficult to mine, and more accurate mining of crowd features cannot be performed in a targeted manner are solved.

Description

Background Art

[0002] Minors below the age limit of adults are the potential force for national development. As key personnel, they are the hope of the whole society. However, due to their weak cognitive ability and immature physical and mental development, they are very susceptible to the influence of bad social trends, especially with the development of modern media and various ways of making friends. Once they make friends with bad people, they are easily affected by bad influences and even become corrupt and commit crimes, which has led to a gradual increase in the rate of juvenile delinquency in recent years. In-depth analysis of such juvenile delinquency cases and the social relationships of delinquent minors can provide a deeper understanding of the reasons and motivations for juvenile delinquency and propose more effective prevention and intervention measures, which is of great significance for promoting the healthy growth of minors and the harmonious development of society.

[0003] The data sources for relationship analysis of minors in related technologies are relatively single, making it difficult to explore the potential social relationships of key target individuals, and there is no targeted and more precise exploration of target groups and population characteristics. Summary of the invention

[0004] The purpose of this application is to provide a method, device and equipment for analyzing the relationship network of key personnel, which is used to solve the problem that the existing solutions for analyzing the relationship of minors have relatively single source data, it is difficult to mine the potential social relationships of key target personnel, and there is no targeted and more accurate mining of target groups and population characteristics.

[0005] In a first aspect, an embodiment of the present application provides a method for analyzing a key personnel relationship network, the method comprising:

[0006] Based on the basic information of different key personnel in the database, obtain the personnel contact information of different key personnel from different scenario databases;

[0007] Each key person and the close contacts who have contact with the key person are respectively regarded as nodes, and connected by edges, and the degree of close contact between the key person and the close contacts is used as the weight of the edge to obtain a relationship map of key persons;

[0008] Divide the nodes in the key personnel relationship map into multiple communities, measure the degree of close connection between nodes in the community based on the weights of the edges between nodes in the community to obtain the modularity of the community, and optimize the division method of the community based on the modularity until the community update end condition is met;

[0009] The obtained personnel attribute information of the nodes of each community is used as a sample, the number of clusters is determined for the sample set corresponding to each community, and clustering is performed according to the number of clusters;

[0010] After clustering, the characteristic analysis of the personnel attribute information in each cluster is performed.

[0011] The key personnel relationship network analysis method, device and equipment provided in the embodiment of the present application can integrate data from different scenario databases into a unified analysis environment; can be flexibly applied to various scenarios, can mine the relationship network between potential key personnel through individual behavior, deeply understand the relationship between individuals, and discover closely related groups and group characteristics. This solves the problem that the existing solutions for analyzing the relationship between minors have relatively single source data, making it difficult to mine the potential social relationships of target key personnel, and there is no targeted and more accurate mining of target groups and population characteristics.

[0012] In some possible embodiments, before dividing the nodes in the key personnel relationship graph into a plurality of communities based on a community discovery algorithm optimized by modularity, the method further includes:

[0013] The edges and the nodes connected thereto whose weights exceed a set threshold in the key personnel relationship map are screened out to obtain an optimized key personnel relationship map.

[0014] The embodiment of the present application retains the edges whose weights exceed a set threshold and the nodes connected to them, thereby retaining the key personnel and close contacts with a high degree of close contact in the key personnel relationship map, and can perform subsequent population division more accurately.

[0015] In some possible embodiments, the method further includes:

[0016] After obtaining the personnel contact information of different key personnel from different scenario databases, for each key personnel, the personnel contact information is sorted into a data format of key personnel ID, close contact personnel ID, number of contacts, and contact scenario; and / or

[0017] For any key person, the degree of close contact between the key person and any close contact is determined as follows: degree of close contact = ∑ (number of contacts × scenario weight factor), where ∑ represents the sum of different scenarios, wherein corresponding scenario weight factors are defined for different scenarios;

[0018] The normalized close degree is obtained by subtracting the minimum close degree from any close degree and dividing it by the difference between the maximum close degree and the minimum close degree.

[0019] The embodiment of the present application can be more conducive to subsequent close contact analysis by organizing personnel contact information into a fixed data format; by assigning different weights to different scenarios and performing weighted summation of the weights for different scenarios, the close contact degree of multiple scenarios can be obtained; by normalizing the close contact degree, it can be organized into a range of 0 to 1 for subsequent community division.

[0020] In some possible embodiments, the nodes in the key personnel relationship map are divided into multiple communities, the degree of close connection between nodes in the community is measured based on the weight of the edges between nodes in the community to obtain the modularity of the community, and the division method of the community is optimized based on the modularity until the community update end condition is met, including:

[0021] Initialize each node in the key personnel relationship graph as a separate community, and trigger a modularity calculation instruction;

[0022] In response to the modularity calculation instruction, the modularity is calculated based on the sum of the weights of all edges in the key personnel relationship graph, the weights of the edges between two nodes, whether the two nodes belong to the same community, and the weights of all edges connected to any node;

[0023] When it is determined that the community update end condition has not been met, the current community is updated by moving the node to an adjacent community, and the modularity calculation instruction is triggered to recalculate the modularity, and the community update end condition is that the modularity reaches a maximum;

[0024] When it is determined that the community update end condition is met, the update of the community is ended.

[0025] The embodiment of the present application calculates the modularity based on the sum of the weights of all edges, the weights of the edges between any two nodes, whether any two nodes belong to the same community, and the weights of all edges connected to any node for the key personnel relationship graphs obtained in different scenarios, and proposes a method for community division based on the principle of maximizing modularity, so that the degree of close contact between people in the community is relatively high, while the degree of contact between communities is relatively sparse, thereby achieving a preliminary division of the population.

[0026] In some possible embodiments, the modularity is calculated based on the sum of the weights of all edges in the key personnel relationship graph, the weights of the edges between two nodes, whether the two nodes belong to the same community, and the weights of all edges connected to any node, including:

[0027] Calculate modularity:

[0028] Among them, m is the sum of the weights of all edges in the key personnel relationship graph, A ij is the weight of the edge between node i and node j. If there is no edge between node i and node j, A ij is 0, k i is the sum of the weights of all edges connected to node i, k j is the sum of the weights of all edges connected to node j, δ(c i c j ) is the indicator function. If the community c to which node i belongs i and the community c to which node j belongs jWhen they are in the same community, the indicator function is 1; otherwise, the indicator function is 0.

[0029] The modularity calculation method proposed in the embodiment of the present application, which is based on the sum of the weights of all edges, the weights of the edges between two nodes, whether two nodes belong to the same community, and the weights of all edges connected to any node, can more accurately measure the degree of close connection between nodes in the community.

[0030] In some possible embodiments, determining the number of clusters for the sample sets corresponding to each community, and performing clustering according to the number of clusters includes:

[0031] Initialize the number of clusters k and trigger the sample clustering instruction;

[0032] In response to the sample clustering instruction, k samples are randomly selected from the samples corresponding to the community as the initial clustering centers;

[0033] Calculate the distance from each sample to each cluster center, assign the sample to the cluster to which the nearest cluster center belongs, and determine whether the clustering end condition is met;

[0034] When it is determined that the clustering end condition is not met, the cluster center of each cluster and the distance from each sample to each cluster center are recalculated based on the samples in the current cluster, the sample is assigned to the cluster to which the cluster center closest to it belongs, and it is determined whether the clustering end condition is met;

[0035] When it is determined that the clustering end condition is met, the sum of squared errors (SSE) is calculated based on the distance between the samples in each cluster and the cluster center. The clustering end condition includes that the cluster center does not change or reaches a preset number of iterations;

[0036] When it is determined according to the SSE that the k value update termination condition is not met, the cluster number k is increased and the sample clustering instruction is triggered, otherwise the sample clustering is terminated, and the k value update termination condition includes that the decrease amplitude of the SSE is less than the set amplitude.

[0037] The embodiment of the present application adopts the above-mentioned elbow method to determine clustering. The core idea of ​​the elbow method is: as the number of clusters k increases, the sample division will be more refined, and the degree of aggregation of each cluster will gradually increase, so the error square sum SSE will naturally gradually decrease. In addition, when k is less than the actual number of clusters, since the increase in k will greatly increase the degree of aggregation of each cluster, the decrease in SSE will be large, and when k reaches the actual number of clusters, the return of the degree of aggregation obtained by increasing k will rapidly decrease, so the decrease in SSE will drop sharply, and then tend to be flat as the k value continues to increase. That is to say, the relationship graph between SSE and k is in the shape of an elbow, and the k value corresponding to this elbow is the actual number of clusters of the data, so the population can be clustered more accurately.

[0038] In some possible embodiments, the sum of squared errors SSE is calculated based on the distance between the samples in each cluster and the cluster center:

[0039]

[0040] Among them, C i is the i-th cluster, p is C i The sample in m i It is C i The mean of all samples in the cluster.

[0041] In the embodiment of the present application, the SSE is determined by the above method, and whether the real clustering number k value is reached can be measured by SSE to support subsequent more accurate population clustering.

[0042] In a second aspect, an embodiment of the present application provides a key personnel relationship network analysis device, the device comprising:

[0043] A contact information acquisition module is used to obtain the personnel contact information of different key personnel from different scenario databases based on the basic information of different key personnel in the database;

[0044] A relationship map determination module is used to respectively regard each key person and the close contacts who have contact with the key person as nodes, and connect them with edges, and use the degree of close contact between the key person and the close contacts as the weight of the edge to obtain a relationship map of the key persons;

[0045] A community division module is used to divide the nodes in the key personnel relationship map into multiple communities, measure the degree of close connection between nodes in the community based on the weight of the edges between nodes in the community to obtain the modularity of the community, and optimize the division method of the community based on the modularity until the community update end condition is met;

[0046] A sample clustering module, used to use the obtained personnel attribute information of the nodes of each community as samples, determine the number of clusters for the sample set corresponding to each community, and perform clustering according to the number of clusters;

[0047] The feature analysis module is used to perform feature analysis on the personnel attribute information in each cluster after clustering is completed.

[0048] In a third aspect, an embodiment of the present application provides a key personnel relationship network analysis device, comprising at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method provided in the first aspect above.

[0049] In a fourth aspect, an embodiment of the present application provides a computer storage medium, wherein the computer storage medium stores a computer program, and the computer program is used to enable a computer to execute the method provided in the first aspect above.

[0050] Other features and advantages of the present application will be described in the following description, and partly become apparent from the description, or be understood by practicing the present application. The purpose and other advantages of the present application can be realized and obtained by the structures specifically pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application will be briefly introduced below. Obviously, the drawings introduced below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0052] Figure 1 A schematic diagram of a software framework for cloud services in the related art;

[0053] Figure 2 A schematic diagram of a method for analyzing a key personnel relationship network according to an embodiment of the present application;

[0054] Figure 3 A schematic diagram of a key personnel relationship map constructed according to an embodiment of the present application;

[0055] Figure 4 A schematic diagram of processing collected personnel contact information to obtain a close contact degree according to an embodiment of the present application;

[0056] Figure 5 A schematic diagram of discovering community groups using the Louvain algorithm according to an embodiment of the present application;

[0057] Figure 6 A detailed flow chart of clustering a sample set corresponding to a community using a clustering algorithm according to an embodiment of the present application;

[0058] Figure 7 A schematic diagram of a key personnel relationship network analysis device according to an embodiment of the present application;

[0059] Figure 8 Schematic diagram of a key personnel relationship network analysis device according to an embodiment of the present application. DETAILED DESCRIPTION

[0060] To further illustrate the technical solution provided by the embodiment of the present application, this is described in detail below in conjunction with the accompanying drawings and specific implementation methods. Although the embodiment of the present application provides the method operation steps as shown in the following embodiments or drawings, more or fewer operation steps may be included in the method based on routine or no creative labor. In the steps where there is no necessary causal relationship logically, the execution order of these steps is not limited to the execution order provided by the embodiment of the present application. The method can be executed in the order of the method shown in the embodiment or drawings or in parallel during the actual processing process or when the control device is executed.

[0061] The technical terms involved in this application are explained and illustrated below.

[0062] 1) Knowledge Graph

[0063] A graph-based data structure consisting of nodes and edges, where each node represents an "entity" and each edge represents the "relationship" between entities. The knowledge graph is essentially a semantic network that uses visualization technology to describe knowledge resources and their carriers, and mine, analyze, construct, draw and display knowledge and their connections. It displays complex knowledge fields through data mining, information processing, knowledge measurement and graphics drawing, reveals the dynamic development laws of knowledge fields, and provides practical references for subject research.

[0064] 2) Community Discovery Algorithm

[0065] Community detection is a common social network analysis task, which aims to automatically identify closely interconnected subnetworks in a social network, which are called communities.

[0066] 3) Data collection

[0067] Data collection, also known as data acquisition, is an interface that uses a device to collect data from outside the system and input it into the system. Data collection technology is widely used in various fields.

[0068] 4) neo4j

[0069] A high-performance NOSQL graph database that stores structured data on the network instead of in tables. It is an embedded, disk-based, fully transactional Java persistence engine, but it stores structured data on the network (called a graph from a mathematical point of view) instead of in tables.

[0070] 5) Louvain algorithm

[0071] A community discovery algorithm based on modularity optimization aims to divide the nodes in the network into different communities, so that the nodes within each community are closely connected, while the connections between different communities are relatively sparse. The algorithm measures the quality of community division by maximizing modularity, where modularity is a measure of the influence of nodes in the network on their neighbors.

[0072] 6) Cloud Services

[0073] Cloud service refers to a server cluster that uses virtualized cloud computing technologies to provide users with a convenient, pay-as-you-go Internet service through the Internet. Cloud services provide great flexibility in resource allocation, replication, and expansion to balance the needs of users, hosted applications, and solutions. Cloud services are built, operated, and managed by cloud service providers, who are committed to ensuring the end-to-end availability, reliability, and security of the cloud.

[0074] The related technology provides a method for obtaining multi-dimensional personnel relationships and event associations, including obtaining information data of target detainees from public security supervision big data; extracting personnel who have event associations or relative associations with the target detainees from the information data of the target detainees to create a personnel relationship model, and at the same time creating an event association model based on the event and personnel relationship; according to the created personnel relationship model or event association model, setting warning values ​​for the associated personnel or events, and issuing warnings for personnel or events that exceed the warning values. The advantages of the present invention are: not only can it realize a comprehensive and integrated multi-dimensional analysis of the target detainees, and deeply excavate various potential relationships hidden behind the target detainees, thereby realizing a more accurate warning function and ensuring prison safety; but it can also provide new ideas for the design of personnel file management system software.

[0075] The existing technology has three main defects:

[0076] First, it is difficult to explore the potential social relationships of the target persons, as the source data is relatively single and can only monitor the potential relationships between the target detainees;

[0077] Second, it only analyzes the relationships between the target detainees, without conducting more precise research on the target groups and their characteristics;

[0078] Third, the accuracy of using similarity algorithms to infer the relationship between people is relatively low, and the computing resources and time costs are high.

[0079] In view of the above-mentioned problems existing in the related technology, the present application proposes a method, device and equipment for analyzing the relationship network of key personnel.

[0080] Other features and advantages of the present application will be described in the following description, and partly become apparent from the description, or be understood by practicing the present application. The purpose and other advantages of the present application can be realized and obtained by the structures specifically pointed out in the written description, claims, and drawings.

[0081] Cloud service refers to a server cluster that uses virtualized cloud computing technologies to provide users with a convenient, pay-as-you-go Internet service through the Internet. Cloud services provide great flexibility in resource allocation, replication, and expansion to balance the needs of users, hosted applications, and solutions. Cloud services are built, operated, and managed by cloud service providers, who are committed to ensuring the end-to-end availability, reliability, and security of the cloud.

[0082] Cloud services are divided according to service models, such as Figure 1 As shown in the figure, the mainstream is divided into the following three layers: Software as a Service SaaS, Platform as a Service PaaS and Infrastructure as a Service IaaS. The three-layer architecture is both independent and closely coordinated with each other, providing powerful parallel computing and distributed deployment capabilities for cloud services.

[0083] This application relies on cloud services to achieve efficient data collection, storage, processing and analysis. It is a model that spans the PaaS big data center layer and the SaaS data analysis layer. The PaaS layer big data components (Hadoop, spark, Docker, KubernetesPod, etc.) provide capability support and performance guarantee for the upper layer SaaS.

[0084] Deploy the components required for big data processing on the cloud platform, such as distributed computing frameworks such as Hadoop and Spark, as well as tools and services such as data warehouses, data mining, and machine learning. These components and tools will together constitute a big data processing platform to meet the standardized, streamlined, and automated data governance and analysis service needs such as big data access, data governance, data analysis, and data services. In the present invention, cloud service big data processing technology is used for early data set preparation and for data comparison and analysis models.

[0085] Deploy the graph database (Neo4j library) and required components (Docker) on the cloud platform. After the deployment is successful, you can use Neo4j's graph query language Cypher to perform addition, deletion, modification, query and analysis tasks to achieve in-depth mining of graph knowledge associations. This application embodiment provides a key personnel relationship network analysis method based on cloud services, such as Figure 2 As shown, it mainly includes:

[0086] Step 201, based on the basic information of different key personnel in the database, obtaining personnel contact information of different key personnel from different scenario databases;

[0087] The key personnel in the embodiments of the present application may be, but are not limited to, delinquent minors. The data of relevant illegal management departments are gathered on the cloud service big data platform, so that the basic information of the key personnel can be obtained through the cloud service big data platform. By way of example, it may include basic information of delinquent minors, case information, and travel information of illegal management departments. The information stored in the above-mentioned different scenario databases is specifically registration information for different scenarios, which may be, but is not limited to, hotel accommodation information, Internet cafe Internet access information, entertainment venue entry and exit information, medical record information, and entry and exit information of other places that require real-name registration.

[0088] The above-mentioned scenario databases include relational databases (such as MySQL, PostgreSQL, Oracle) or non-relational databases (such as MongoDB, Redis). After obtaining the basic information of key personnel, the time, place and contact person information that appear in different scenario databases can be determined based on the basic information. Specifically, a variety of data input strategies, data output strategies and data access modes that match the business scenarios can be defined to process the data, including a series of data processing steps such as extraction, cleaning, association, comparison, and identification, to structure the unstructured data and form a data organization with reasonable classification, unified format, and proper configuration.

[0089] The embodiment of the present application is based on the functions of the cloud service big data platform, and performs operations such as cleaning, preprocessing, governance, extraction and standardization on the aggregated data to form a data set. By constructing an association comparison model, the personnel contact information between key personnel is extracted to form an original data set, and data preparation is made for analyzing the close contacts and relationship groups of key personnel. In the embodiment of the present application, the close contacts of key personnel are those who have a contact relationship with the key personnel in at least one scenario. According to the number of contacts, the personnel whose close contact degree exceeds the set threshold are classified as confidential personnel, and the close contacts include key personnel and / or non-key personnel.

[0090] In some possible embodiments, after obtaining the personnel contact information between different key personnel from different scenario databases, the personnel contact information is organized into a data format of key personnel ID, close contact ID, number of contacts, and contact scenario for each key personnel.

[0091] Step 202: each key person and the close contacts who have contact with the key person are respectively regarded as nodes, and connected by edges, and the degree of close contact between the key person and the close contacts is used as the weight of the edge, so as to obtain a relationship map of the key persons;

[0092] The embodiment of the present application constructs a key personnel relationship map based on the personnel contact information of different key personnel, and specifically, the key personnel relationship map can be constructed through a cloud service graph database (neo4j database). Specifically, the personnel contact information structure is imported into the cloud service graph database as a triple, and the information of the above triple includes: node definition: personnel ID (key personnel ID / close contact personnel ID); weighted edge definition: close contact degree, node attribute type: attribute information of key personnel / close contact personnel (name, ID number, case type, violation record and other attribute information), forming a cloud service key personnel relationship map.

[0093] Neo4j database can efficiently store and query complex user relationships in social networks. It can use users as nodes and contact relationships between users as relationships. Through the direction and hierarchy of relationships, it can intuitively display the social network structure and hierarchical relationships between users. The basic information of key individuals (including name, ID number, gender, education background, home address, contact information, family background, case type, violation record, etc.) is stored in the graph database as node attribute information, which can quickly and conveniently display the basic information of individuals when querying.

[0094] like Figure 3 A relationship map of key personnel in an example of an embodiment of the present application is shown.

[0095] Step 203, dividing the nodes in the key personnel relationship map into multiple communities, measuring the degree of close connection between nodes in the community based on the weights of the edges between nodes in the community to obtain the modularity of the community, and optimizing the division method of the community based on the modularity until the community update end condition is met;

[0096] Based on the constructed relationship map of key cloud service personnel, the Louvain algorithm is used to divide the community. The modularity of the community is obtained by measuring the degree of connection between nodes in the community based on the weight of the edges between nodes in the community, so that the internal nodes of the divided communities are closely connected, while the connections between different communities are relatively sparse. After obtaining the modularity, the community division method is continuously optimized on the principle of maximizing the modularity until the community update conditions are met. The community update conditions can be that the optimization times reach a certain number or that the modularity meets the requirements.

[0097] Step 204, using the obtained personnel attribute information of the nodes of each community as samples, determining the number of clusters for the sample set corresponding to each community, and performing clustering according to the number of clusters;

[0098] Step 205: After the clustering is completed, feature analysis is performed on the personnel attribute information in each cluster.

[0099] The key personnel relationship network analysis method, device and equipment provided in the embodiment of the present application can integrate data from different scenario databases into a unified analysis environment; can be flexibly applied to various scenarios, can mine the relationship network between potential key personnel through individual behavior, deeply understand the relationship between individuals, and discover closely related groups and group characteristics. This solves the problem that the existing solutions for analyzing the relationship between minors have relatively single source data, making it difficult to mine the potential social relationships of target key personnel, and there is no targeted and more accurate mining of target groups and population characteristics.

[0100] In the embodiment of the present application, corresponding weights are defined for different scenarios. For any key person, the degree of close contact between the key person and any close contact person is determined as follows: degree of close contact = ∑ (number of contacts × scenario weight factor), where ∑ represents the sum of different scenarios, wherein corresponding scenario weight factors are defined for different scenarios; illustratively, the weights defined for different scenarios are:

[0101] Same case: 20; staying in the same hotel: 10; surfing the Internet in the same Internet cafe: 5; going to and from the same entertainment venue: 3; seeking medical treatment together: 2; traveling to the same illegal management department: 1.

[0102] Furthermore, the minimum value of the closeness is subtracted from any closeness, and the result is divided by the difference between the maximum value of the closeness and the minimum value of the closeness to obtain the normalized closeness. The scale of the closeness is scaled to between 0 and 1, which can be specifically expressed as follows:

[0103]

[0104] min(x) is the minimum value of the close connection degree in the data set composed of all close connection degrees, and max(x) is the maximum value of the close connection degree in the data set.

[0105] In some possible embodiments, before dividing the nodes in the key personnel relationship graph into a plurality of communities, the method further includes:

[0106] The edges and the nodes connected thereto whose weights exceed a set threshold in the key personnel relationship map are screened out to obtain an optimized key personnel relationship map.

[0107] like Figure 4As shown is a framework diagram of an embodiment of the present application, the data collection part is responsible for collecting the following data: basic information of minors who violate the law, travel information of the violation management department, Internet cafe registration information, hotel accommodation registration information, entertainment venue entry and exit information, and other venues that require real-name registration; the same case, same room, same accommodation, same Internet access, same flight, same bus, same train, same rental house, same entry and exit of entertainment venues, etc. are used as contact information, and the degree of close contact between any key person and different close contacts is determined according to the above method provided in the embodiment of the present application, the close contact degree is normalized to 0-1, and rating is performed according to the close contact degree score.

[0108] Exemplarily, the close contact degree is divided into: 0-0.4: very low; 0.4-0.6: low; 0.6-0.8: high; 0.8-1: very high, so that a data set consisting of an array of key personnel ID, close contact personnel ID, and close contact degree can be obtained.

[0109] In some possible embodiments, the nodes in the key personnel relationship map are divided into multiple communities, the degree of close connection between nodes in the community is measured based on the weight of the edges between nodes in the community to obtain the modularity of the community, and the division method of the community is optimized based on the modularity until the community update end condition is met, including:

[0110] Initialize each node in the key personnel relationship graph as a separate community, and trigger a modularity calculation instruction;

[0111] In response to the modularity calculation instruction, the modularity is calculated based on the sum of the weights of all edges in the key personnel relationship map, the weights of the edges between two nodes, whether the two nodes belong to the same community, and the weights of all edges connected to any node. The modularity represents the community division instruction. The larger the modularity, the tighter the node connections within the community, and the greater the relative coefficient of connection between different communities.

[0112] When it is determined that the community update end condition has not been met, the current community is updated by moving the node to an adjacent community, and the modularity calculation instruction is triggered to recalculate the modularity, and the community update end condition is that the modularity reaches a maximum;

[0113] When it is determined that the community update end condition is met, the update of the community is ended.

[0114] In some possible embodiments, the modularity is calculated based on the sum of the weights of all edges in the key personnel relationship graph, the weights of the edges between two nodes, whether the two nodes belong to the same community, and the weights of all edges connected to any node, including:

[0115] Calculate modularity:

[0116] Among them, m is the sum of the weights of all edges in the key personnel relationship graph, A ij is the weight of the edge between node i and node j. If there is no edge between node i and node j, A ij is 0, k i is the sum of the weights of all edges connected to node i, k j is the sum of the weights of all edges connected to node j, δ(c i c j ) is the indicator function. If the community c to which node i belongs i and the community c to which node j belongs j When they are in the same community, the indicator function is 1; otherwise, the indicator function is 0.

[0117] The following is a detailed process of dividing the nodes in the key personnel relationship map into multiple communities in the community discovery algorithm based on modularity optimization in the embodiment of the present application with reference to the accompanying drawings. Figure 5 As shown in the figure, after obtaining the relationship graph of key personnel and filtering the data by selecting nodes whose edge weights are greater than a set threshold such as 0.6, the flowchart of using the Louvain algorithm to divide the community mainly includes:

[0118] The first step is to initialize each node in the key personnel relationship graph as a separate community;

[0119] The second step is to calculate the modularity using the modularity calculation method provided in the embodiment of the present application;

[0120] In the third step, for each node, an algorithm is used to try to move the node to an adjacent community to see if the modularity can be increased.

[0121] In the fourth step, the second and fourth steps are iterated until the node movement no longer increases the modularity. In each iteration, the algorithm re-evaluates the edge weights to reflect the changes in the network structure after the node movement.

[0122] The fifth step is to output the community to which each node belongs to form the community division result.

[0123] The above community division results are further analyzed. A sample set is formed based on the attribute information of each person in the community, including age, gender, education, occupation, place of residence, family background (single-parent / orphan / left-behind), criminal record, frequent Internet access, frequent hotel stays, etc. The key personnel in the sample set are classified through the K-means clustering algorithm to further explore the basic characteristics of the social relations of the illegal juvenile population.

[0124] In some possible embodiments, the number of clusters is determined for the sample set corresponding to each community. Specifically, the number of clusters may be determined by using an elbow method, and clustering is performed according to the number of clusters, including:

[0125] Initialize the number of clusters k and trigger the sample clustering instruction;

[0126] In response to the sample clustering instruction, k samples are randomly selected from the samples corresponding to the community as the initial clustering centers;

[0127] Calculate the distance from each sample to each cluster center, assign the sample to the cluster to which the nearest cluster center belongs, and determine whether the clustering end condition is met;

[0128] When it is determined that the clustering end condition is not met, the cluster center of each cluster and the distance from each sample to each cluster center are recalculated based on the samples in the current cluster, the sample is assigned to the cluster to which the cluster center closest to it belongs, and it is determined whether the clustering end condition is met;

[0129] When it is determined that the clustering end condition is met, the sum of squared errors (SSE) is calculated based on the distance between the samples in each cluster and the cluster center. The clustering end condition includes that the cluster center does not change or reaches a preset number of iterations;

[0130] When it is determined according to the SSE that the k value update termination condition is not met, the cluster number k is increased and the sample clustering instruction is triggered, otherwise the sample clustering is terminated, and the k value update termination condition includes that the decrease amplitude of the SSE is less than the set amplitude.

[0131] SSE is the clustering error of all samples, which represents the quality of clustering effect. The core idea of ​​the elbow method is: as the number of clusters k increases, the sample division will be more refined, and the degree of aggregation of each cluster will gradually increase, so the error square sum SSE will naturally gradually decrease. In addition, when k is less than the actual number of clusters, since the increase in k will greatly increase the degree of aggregation of each cluster, the decrease in SSE will be large. When k reaches the actual number of clusters, the return of the degree of aggregation obtained by increasing k will rapidly decrease, so the decrease in SSE will drop sharply, and then tend to be flat as the k value continues to increase. In other words, the relationship between SSE and k is in the shape of an elbow, and the k value corresponding to this elbow is the actual number of clusters of the data.

[0132] In some possible embodiments, the sum of squared errors SSE is calculated based on the distance between the samples in each cluster and the cluster center:

[0133]

[0134] Among them, C i is the i-th cluster, p is C i The sample in m i It is C i The mean of all samples in the cluster.

[0135] like Figure 6 The specific process of sample clustering in the embodiment of the present application is shown, which specifically includes:

[0136] Step 601, initializing the number of clusters k;

[0137] Step 602, selecting k samples from a sample set of N samples as initial cluster centers;

[0138] Step 603, for the current cluster center, calculate the distance between each sample and each cluster center, and assign the sample to the cluster to which the cluster center with the closest distance belongs;

[0139] Using the Euclidean distance formula C (i) :=arg min||x (i) -μ j ||, for each sample x i According to the principle of the shortest distance, the sample data set is clustered to determine the class relationship of the samples, μ j is the cluster center of the jth cluster, specifically the mean of all samples in the jth cluster. i Represents sample x i The cluster that is closest to the k clusters.

[0140] Step 604, recalculating the cluster center of each cluster based on the samples in the current cluster, usually calculating the mean of all sample points in the cluster;

[0141] Step 605, repeating steps 603 and 604 until the cluster center no longer changes or a preset number of iterations is reached;

[0142] Step 606, calculating the sum of squared errors (SSE) according to the distance between the samples in each cluster and the centroid of the cluster;

[0143] Step 607, determining whether the SSE decrease amplitude is less than the set amplitude, if yes, execute step 608, otherwise execute step 609;

[0144] Step 608, increase the k value and return to step 602;

[0145] Step 609, end sample clustering, and obtain the population classification results and population proportions.

[0146] Based on the same inventive concept, the present application also provides a key personnel relationship network analysis device, such as Figure 7 As shown, the device comprises:

[0147] The contact information acquisition module 701 is used to acquire the personnel contact information of different key personnel from different scenario databases based on the basic information of different key personnel in the database;

[0148] The relationship map determination module 702 is used to respectively regard each key person and the close contacts who have contact with the key person as nodes, and connect them with edges, and regard the degree of close contact between the key person and the close contacts as the weight of the edge, so as to obtain a relationship map of the key persons;

[0149] The community division module 703 is used to divide the nodes in the key personnel relationship map into multiple communities, measure the close connection between the nodes in the community based on the weight of the edges between the nodes in the community to obtain the modularity of the community, and optimize the division method of the community based on the modularity until the community update end condition is met;

[0150] The sample clustering module 704 is used to use the obtained personnel attribute information of the nodes of each community as samples, determine the number of clusters for the sample set corresponding to each community, and perform clustering according to the number of clusters;

[0151] The feature analysis module 705 is used to perform feature analysis on the personnel attribute information in each cluster after the clustering is completed.

[0152] In some possible embodiments, before the community division module divides the nodes in the key personnel relationship graph into a plurality of communities based on a community discovery algorithm optimized for modularity, the community division module further includes:

[0153] The edges and the nodes connected thereto whose weights exceed a set threshold in the key personnel relationship map are screened out to obtain an optimized key personnel relationship map.

[0154] In some possible embodiments, the device further includes:

[0155] A data preprocessing module is used to obtain the personnel contact information of different key personnel from different scenario databases, and organize the personnel contact information into the data format of key personnel ID, close contact personnel ID, number of contacts, and contact scenario for each key personnel; and / or

[0156] A close contact degree determination module is used to determine the close contact degree between any key person and any close contact person as follows: close contact degree = ∑ (number of contacts × scenario weight factor), ∑ represents the sum of different scenarios, wherein corresponding scenario weight factors are defined for different scenarios;

[0157] The close contact degree normalization module is used to subtract the minimum close contact degree from any close contact degree and divide it by the difference between the maximum close contact degree and the minimum close contact degree to obtain the normalized close contact degree.

[0158] In some possible embodiments, the community division module divides the nodes in the key personnel relationship map into multiple communities, measures the degree of close connection between nodes in the community based on the weights of the edges between nodes in the community to obtain the modularity of the community, and optimizes the division method of the community based on the modularity until the community update end condition is met, including:

[0159] Initialize each node in the key personnel relationship graph as a separate community, and trigger a modularity calculation instruction;

[0160] In response to the modularity calculation instruction, the modularity is calculated based on the sum of the weights of all edges in the key personnel relationship graph, the weights of the edges between two nodes, whether the two nodes belong to the same community, and the weights of all edges connected to any node;

[0161] When it is determined that the community update end condition has not been met, the current community is updated by moving the node to an adjacent community, and the modularity calculation instruction is triggered to recalculate the modularity, and the community update end condition is that the modularity reaches a maximum;

[0162] When it is determined that the community update end condition is met, the update of the community is ended.

[0163] In some possible embodiments, the community division module calculates the modularity based on the sum of the weights of all edges in the key personnel relationship graph, the weights of the edges between two nodes, whether the two nodes belong to the same community, and the weights of all edges connected to any node, including:

[0164] Calculate modularity:

[0165] Among them, m is the sum of the weights of all edges in the key personnel relationship graph, A ij is the weight of the edge between node i and node j. If there is no edge between node i and node j, A ij is 0, k i is the sum of the weights of all edges connected to node i, k j is the sum of the weights of all edges connected to node j, δ(c i c j ) is the indicator function. If the community c to which node i belongs i and the community c to which node j belongs j When they are in the same community, the indicator function is 1; otherwise, the indicator function is 0.

[0166] In some possible embodiments, the sample clustering module uses the elbow method to determine the number of clusters for the sample set corresponding to each community, and performs clustering according to the number of clusters, including:

[0167] Initialize the number of clusters k and trigger the sample clustering instruction;

[0168] In response to the sample clustering instruction, k samples are randomly selected from the samples corresponding to the community as the initial clustering centers;

[0169] Calculate the distance from each sample to each cluster center, assign the sample to the cluster to which the nearest cluster center belongs, and determine whether the clustering end condition is met;

[0170] When it is determined that the clustering end condition is not met, the cluster center of each cluster and the distance from each sample to each cluster center are recalculated based on the samples in the current cluster, the sample is assigned to the cluster to which the cluster center closest to it belongs, and it is determined whether the clustering end condition is met;

[0171] When it is determined that the clustering end condition is met, the sum of squared errors (SSE) is calculated based on the distance between the samples in each cluster and the cluster center. The clustering end condition includes that the cluster center does not change or reaches a preset number of iterations;

[0172] When it is determined according to the SSE that the k value update termination condition is not met, the cluster number k is increased and the sample clustering instruction is triggered, otherwise the sample clustering is terminated, and the k value update termination condition includes that the decrease amplitude of the SSE is less than the set amplitude.

[0173] In some possible embodiments, the sample clustering module calculates the sum of squared errors SSE according to the distance between the samples in each cluster and the cluster center:

[0174]

[0175] Among them, C i is the i-th cluster, p is C i The sample in m i It is C i The mean of all samples in the cluster.

[0176] After introducing the key personnel relationship network analysis method and apparatus according to the exemplary embodiment of the present application, next, a key personnel relationship network analysis device according to another exemplary embodiment of the present application is introduced.

[0177] Those skilled in the art will appreciate that various aspects of the present application may be implemented as a system, method or program product. Therefore, various aspects of the present application may be specifically implemented in the following forms, namely: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software, which may be collectively referred to as "circuit", "module" or "system" herein.

[0178] In some possible implementations, the key personnel relationship network analysis device according to the present application may include at least one processor and at least one memory. The memory stores program code, and when the program code is executed by the processor, the processor executes the steps of the key personnel relationship network analysis method according to various exemplary implementations of the present application described above in this specification.

[0179] Refer to the following Figure 8 To describe the key personnel relationship network analysis device 180 according to this embodiment of the present application. Figure 8 The key personnel relationship network analysis device 180 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0180] like Figure 8 As shown, the key personnel relationship network analysis device 180 is in the form of a general electronic device. The components of the key personnel relationship network analysis device 180 may include but are not limited to: at least one processor 181, at least one memory 182, and a bus 183 connecting different system components (including the memory 182 and the processor 181).

[0181] Bus 183 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, a processor, or a local bus using any of a variety of bus architectures.

[0182] The memory 182 may include a readable medium in the form of a volatile memory, such as a random access memory (RAM) 1821 and / or a cache memory 1822 , and may further include a read-only memory (ROM) 1823 .

[0183] The memory 182 may also include a program / utility 1825 having a set (at least one) of program modules 1824, such program modules 1824 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0184] The key personnel relationship network analysis device 180 may also communicate with one or more external devices 184 (e.g., keyboards, pointing devices, etc.), and may also communicate with one or more devices that enable a user to interact with the key personnel relationship network analysis device 180, and / or communicate with any device that enables the key personnel relationship network analysis device 180 to communicate with one or more other electronic devices (e.g., routers, modems, etc.). Such communication may be performed through an input / output (I / O) interface 185. Furthermore, the key personnel relationship network analysis device 180 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through a network adapter 186. Figure 8 As shown, the network adapter 186 communicates with other modules used in the key personnel relationship network analysis device 180 through the bus 183. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with the key personnel relationship network analysis device 180, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0185] In some possible implementations, various aspects of a method for analyzing a key personnel relationship network provided by the present application may also be implemented in the form of a program product, which includes program code. When the program product runs on a computer device, the program code is used to enable the computer device to execute the steps of a method for analyzing a key personnel relationship network according to various exemplary implementations of the present application described above in this specification.

[0186] The program product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0187] The program product for analyzing the relationship network of key personnel of the embodiment of the present application can adopt a portable compact disk read-only memory (CD-ROM) and include program code, and can be run on electronic devices. However, the program product of the present application is not limited to this. In this document, the readable storage medium can be any tangible medium containing or storing a program, which can be used by or in combination with an instruction execution system, an apparatus or a device.

[0188] The readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, wherein the readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The readable signal medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0189] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing.

[0190] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0191] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for analyzing a key personnel relationship network, characterized in that: The method comprises: Based on the basic information of different key personnel in the database, obtain the personnel contact information of different key personnel from different scenario databases; Each key person and the close contacts who have contact with the key person are respectively regarded as nodes, and connected by edges, and the degree of close contact between the key person and the close contacts is used as the weight of the edge to obtain a relationship map of key persons; Divide the nodes in the key personnel relationship map into multiple communities, measure the degree of close connection between nodes in the community based on the weights of the edges between nodes in the community to obtain the modularity of the community, and optimize the division method of the community based on the modularity until the community update end condition is met; The obtained personnel attribute information of the nodes of each community is used as a sample, the number of clusters is determined for the sample set corresponding to each community, and clustering is performed according to the number of clusters; After clustering, the characteristic analysis of the personnel attribute information in each cluster is performed.

2. The method according to claim 1, characterized in that Before dividing the nodes in the key personnel relationship map into a plurality of communities, the method further includes: The edges and the nodes connected thereto whose weights exceed a set threshold in the key personnel relationship map are screened out to obtain an optimized key personnel relationship map.

3. The method according to claim 1, characterized in that Also includes: After obtaining the personnel contact information of different key personnel from different scenario databases, for each key personnel, the personnel contact information is sorted into a data format of key personnel ID, close contact personnel ID, number of contacts, and contact scenario; and / or For any key person, the degree of close contact between the key person and any close contact is determined as follows: degree of close contact = ∑ (number of contacts × scenario weight factor), where ∑ represents the sum of different scenarios, wherein corresponding scenario weight factors are defined for different scenarios; The normalized close degree is obtained by subtracting the minimum close degree from any close degree and dividing it by the difference between the maximum close degree and the minimum close degree.

4. The method according to claim 1, characterized in that: The nodes in the key personnel relationship map are divided into multiple communities, the degree of close connection between the nodes in the community is measured based on the weight of the edges between the nodes in the community to obtain the modularity of the community, and the division method of the community is optimized based on the modularity until the community update end condition is met, including: Initialize each node in the key personnel relationship graph as a separate community, and trigger a modularity calculation instruction; In response to the modularity calculation instruction, the modularity is calculated based on the sum of the weights of all edges in the key personnel relationship graph, the weights of the edges between two nodes, whether the two nodes belong to the same community, and the weights of all edges connected to any node; When it is determined that the community update end condition has not been met, the current community is updated by moving the node to an adjacent community, and the modularity calculation instruction is triggered to recalculate the modularity, and the community update end condition is that the modularity reaches a maximum; When it is determined that the community update end condition is met, the update of the community is ended.

5. The method according to claim 4, characterized in that The modularity is calculated based on the sum of the weights of all edges in the key personnel relationship graph, the weights of the edges between two nodes, whether the two nodes belong to the same community, and the weights of all edges connected to any node, including: Calculate modularity: Among them, m is the sum of the weights of all edges in the key personnel relationship graph, A ij is the weight of the edge between node i and node j. If there is no edge between node i and node j, A ij is 0, k i is the sum of the weights of all edges connected to node i, k j is the sum of the weights of all edges connected to node j, δ(c i c j ) is the indicator function. If the community c to which node i belongs i and the community c to which node j belongs j When they are in the same community, the indicator function is 1; otherwise, the indicator function is 0.

6. The method according to claim 1, characterized in that Determining the number of clusters for the sample set corresponding to each community, and clustering according to the number of clusters, including: Initialize the number of clusters k and trigger the sample clustering instruction; In response to the sample clustering instruction, k samples are randomly selected from the samples corresponding to the community as the initial clustering centers; Calculate the distance from each sample to each cluster center, assign the sample to the cluster to which the nearest cluster center belongs, and determine whether the clustering end condition is met; When it is determined that the clustering end condition is not met, the cluster center of each cluster and the distance from each sample to each cluster center are recalculated based on the samples in the current cluster, the sample is assigned to the cluster to which the cluster center closest to it belongs, and it is determined whether the clustering end condition is met; When it is determined that the clustering end condition is met, the sum of squared errors (SSE) is calculated based on the distance between the samples in each cluster and the cluster center. The clustering end condition includes that the cluster center does not change or reaches a preset number of iterations; When it is determined according to the SSE that the k value update termination condition is not met, the cluster number k is increased and the sample clustering instruction is triggered, otherwise the sample clustering is terminated, and the k value update termination condition includes that the decrease amplitude of the SSE is less than the set amplitude.

7. The method according to claim 6, characterized in that According to the distance between the samples in each cluster and the cluster center, the error sum of squares SSE is calculated: Among them, C i is the i-th cluster, p is C i The sample in m i It is C i The mean of all samples in the cluster.

8. A key personnel relationship network analysis device, characterized in that: The device comprises: A contact information acquisition module is used to obtain the personnel contact information of different key personnel from different scenario databases based on the basic information of different key personnel in the database; A relationship map determination module is used to respectively regard each key person and the close contacts who have contact with the key person as nodes, and connect them with edges, and use the degree of close contact between the key person and the close contacts as the weight of the edge to obtain a relationship map of the key persons; A community division module is used to divide the nodes in the key personnel relationship map into multiple communities, measure the degree of close connection between nodes in the community based on the weight of the edges between nodes in the community to obtain the modularity of the community, and optimize the division method of the community based on the modularity until the community update end condition is met; A sample clustering module, used to use the obtained personnel attribute information of the nodes of each community as samples, determine the number of clusters for the sample set corresponding to each community, and perform clustering according to the number of clusters; The feature analysis module is used to perform feature analysis on the personnel attribute information in each cluster after clustering is completed.

9. A key personnel relationship network analysis device, characterized in that: It comprises at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method as described in any one of claims 1-7.

10. A computer storage medium, characterized in that: The computer storage medium stores a computer program, and the computer program is used to enable a computer to execute the method according to any one of claims 1 to 7.

Images