CN112115213A - Method for guiding evidence-based correction work based on escort personnel relationship network - Google Patents

Abstract

The invention discloses a method for guiding evidence-based correction work based on an escort relationship network. Analyzing the collected social relationship data of the escorting personnel, establishing a corresponding complex network, analyzing the attributes and characteristics of the relationship network of the escorting personnel according to the statistical characteristics, the community discovery algorithm, the information transfer model and the like in the complex network, and researching and calculating the key index of the escorting personnel. The invention finds out the key prisoner by analyzing the status and the action of the escorting personnel in the social network, analyzes the small groups possibly formed by the escorting personnel, can pertinently strengthen the supervision and control, and provides suggestions for reasonable evidence-based correction arrangement.

Description

Translated fromChinese

基于在押人员关系网络指导循证矫治工作的方法A method for guiding evidence-based corrections based on the relationship network of detainees

技术领域technical field

本发明属于复杂网络及其应用领域，主要致力于研究在押人员关系网络中的关键节点，可应用于寻找“关键犯人”，发现监狱中潜在的小团体，指导循证矫治工作等等，具体方法上涉及复杂网络统计分析、网络动力学、社团发现算法和循证矫正等。The invention belongs to the complex network and its application field, and is mainly devoted to the research of key nodes in the relationship network of detainees, and can be applied to find "key prisoners", discover potential small groups in prisons, guide evidence-based correction work, etc. The specific method It involves statistical analysis of complex networks, network dynamics, community discovery algorithms, and evidence-based corrections.

背景技术Background technique

复杂网络理论是近些年迅速崛起的一门新学科，复杂网络分析是研究复杂系统的一种角度和方法，它关注系统中个体相互关联作用的结构，是理解复杂系统性质和功能的一种途径。复杂网络是对复杂系统的一种抽象，例如国际互联网、万维网、电信网络、交通运输网络和人际关系网络等等，都可以建模成一种复杂网络进行分析。Complex network theory is a new discipline that has risen rapidly in recent years. Complex network analysis is a perspective and method of studying complex systems. It focuses on the structure of the interrelated interactions of individuals in the system, and is a way to understand the properties and functions of complex systems. way. A complex network is an abstraction of complex systems, such as the Internet, the World Wide Web, telecommunication networks, transportation networks, and interpersonal networks, etc., which can be modeled as a complex network for analysis.

在一个网络中，不同的节点有不同的地位，起着不同的作用。以微博社交网络为例，有一呼百应的意见领袖，有死寂沉沉的僵尸粉；以交通网络为例，有至关重要的交通枢纽，有无关痛痒的备用中转站。在使用复杂网络分析问题时，如何区分网络中不同节点的重要性程度，是一个需要考虑的重要问题。例如，食物链中哪些物种对整个生态的影响最大？全球经济系统中哪些国家或地区对于体系的有至关重要的作用？“种子信息”发给哪些用户可以带来最大规模的传播？应该采取怎样的免疫策略，避免大规模的疾病传染？应该如何找到“关键犯人”进行监控以避免其煽动他人闹事？复杂网络中的节点重要性为研究此类问题提供了指标。In a network, different nodes have different positions and play different roles. Take the Weibo social network as an example, there are opinion leaders who are always responding, and there are dead zombie fans; taking the transportation network as an example, there are important transportation hubs and innocuous backup transfer stations. When using complex network analysis problems, how to distinguish the importance of different nodes in the network is an important issue that needs to be considered. For example, which species in the food chain have the greatest impact on the entire ecology? Which countries or regions in the global economic system are critical to the system? To which users can "seed information" be sent to bring the greatest spread? What immunization strategies should be adopted to avoid large-scale disease transmission? How to find "key prisoners" for monitoring to avoid inciting others to make trouble? The importance of nodes in complex networks provides indicators for studying such problems.

复杂网络中节点重要性包括度、聚集系数、节点介数中心性等统计指标，这些指标可用于刻画网络的拓扑结构，寻找网络的关键节点或连边。通过建立复杂网络，可以对犯人的一些社会性质作出定量刻画，如犯人和周边犯人联系的广泛度，某个犯人在该关系网络中的地位等。通过找出这些影响较大的“关键犯人”，可以更好地指导狱警的工作。The importance of nodes in complex networks includes statistical indicators such as degree, aggregation coefficient, node betweenness centrality, etc. These indicators can be used to describe the topology of the network and find the key nodes or edges of the network. By establishing a complex network, some social properties of prisoners can be quantitatively described, such as the extent of contact between prisoners and surrounding prisoners, and the status of a prisoner in the relationship network. By identifying these "key prisoners" with greater influence, the work of prison guards can be better guided.

网络动力学即将动力学模型运用到网络中，可以讨论网络上节点的相互影响以及一些不良信息或者行为在这个社会网络上的传递过程。例如通过节点的状态与周围节点状态的关系讨论朋友的行为对于某个个体的影响；可以讨论不良信息在这个社会网络中的传播过程，研究不同节点接触这个信息的过程，在此基础上讨论如何极小化成本及负面影响的控制，包括对某个节点——人进行控制，或者断掉某个边——防止某两个犯人沟通信息。Network dynamics is to apply the dynamic model to the network, which can discuss the mutual influence of nodes on the network and the transmission process of some bad information or behaviors on this social network. For example, the influence of a friend's behavior on an individual can be discussed through the relationship between the state of a node and the state of surrounding nodes; the dissemination process of bad information in this social network can be discussed, and the process of contacting this information by different nodes can be discussed. Controls that minimize costs and negative impacts include controlling a node—person, or breaking an edge—preventing communication between two prisoners.

社团发现又被称为社区检测、图聚类，它是用来揭示网络聚集行为的一种技术。社区检测实际就是一种网络聚类的方法，这里的“社区”并没有一种严格的定义，可以将其理解为一类具有相同特性的节点的集合。一般认为社团内部的点之间的连接相对稠密，而不同社团的点之间的连接相对稀疏。因此社团发现是采用某种方法将网络中的节点进行分组，使得组内节点连接比较紧密，组间节点连接比较稀疏。目前比较成熟的社团发现算法有Fast greedy算法、随机游走算法、层次聚类算法等。通过社团发现，可以寻找潜在的小团体，分析小团体的内在社交关系和关键指数，避免其聚众闹事。Community discovery, also known as community detection and graph clustering, is a technique used to reveal network aggregation behavior. Community detection is actually a method of network clustering. There is no strict definition of "community" here, and it can be understood as a collection of nodes with the same characteristics. It is generally believed that the connections between points within a community are relatively dense, while the connections between points in different communities are relatively sparse. Therefore, community discovery is to use some method to group the nodes in the network, so that the nodes within the group are more closely connected, and the nodes between the groups are more sparsely connected. At present, the more mature community discovery algorithms include Fast greedy algorithm, random walk algorithm, and hierarchical clustering algorithm. Through community discovery, you can find potential small groups, analyze the internal social relations and key indicators of small groups, and avoid crowds and troubles.

循证矫正，本意是“基于证据的矫正”，即针对罪犯的具体问题，结合罪犯的特点和意愿，寻找并按照现有的最佳证据(方法、措施等)来实施矫正活动，其核心是遵循研究证据进行矫正实践，强调罪犯改造的科学性和有效性，从而把研究者的科研成果与矫正工作者的矫正实践结合起来，实现矫正实践的效益最大化。循证矫正是现代科学精神对矫正实践领域的渗透，为罪犯改造工作带来了一场方法论革命。Evidence-based correction, the original meaning is "evidence-based correction", that is, for the specific problems of the offender, combined with the characteristics and wishes of the offender, to find and implement correction activities in accordance with the best available evidence (methods, measures, etc.), the core of which is Follow the research evidence to carry out correction practice, emphasize the scientificity and effectiveness of criminal reform, so as to combine the research results of researchers with the correction practice of correction workers, and maximize the benefits of correction practice. Evidence-based correction is the penetration of modern scientific spirit into the field of correction practice, which brings a methodological revolution to the reform of criminals.

如何安排在押人员询证矫正过程是监狱管理的重要问题。有证据表明，不同的顺序及流程安排会在矫正效果上存在差异，部分原因在于在押人员之间的相互影响所导致。在“监管场所智能监控、预警防范关键技术研发与示范”项目(项目编号：2017YFC0803400)的“多策略异常心理与行为客观判识技术”课题(课题编号：2017YFC0803402)的支持下，我们探索基于在押人员关系网络指导循证矫治工作的方法。以上述方法分析得到的“关键犯人”和小团体的内部社交关系等具体情况为依据，设计合理的矫正活动，并及时跟踪效果并更新矫正活动，以便实现矫正实践的效益最大化。How to arrange the detainee's inquiry and correction process is an important issue of prison management. There is evidence that different sequences and process arrangements have different effects on corrections, in part because of interactions among detainees. With the support of the "Multi-strategy Abnormal Psychology and Behavior Objective Judgment Technology" project (Project No.: 2017YFC0803402) of the "R&D and Demonstration of Key Technologies for Intelligent Monitoring, Early Warning and Prevention in Supervision Sites" (Project No.: 2017YFC0803400), we A method by which the personnel network guides the work of evidence-based corrections. Based on the "key prisoners" and the internal social relations of small groups obtained by the above analysis, reasonable correction activities should be designed, and the effects should be tracked and updated in time, so as to maximize the benefits of correction practices.

发明内容SUMMARY OF THE INVENTION

本发明提出了研究基于在押人员关系网络，利用复杂网络结构和统计性质、网络动力学和社团发现算法，寻找关键节点和潜在小团体的方法。首先将对收集的在押人员社交关系数据进行初步分析整理，建立在押人员关系网络，并进行复杂网络统计特征分析；然后基于该在押人员关系网络，构建信息传递模型，找出网络中的关键节点；通过复杂网络的多种社团发现算法，对在押人员关系网络进行社团划分，寻找潜在的小团体并分析团体内部的相处模式；最后以上述分析的“关键犯人”和小团体为依据，设计适当的的循证矫正顺序和强度，以达到更好的矫正效果，并及时跟踪和更新矫正活动，实现矫正实践的效益最大化。The invention proposes a method for searching for key nodes and potential small groups based on the relationship network of detainees, using complex network structure and statistical properties, network dynamics and community discovery algorithms. Firstly, the collected social relationship data of the detainees will be preliminarily analyzed and sorted, the relationship network of the detainees will be established, and the statistical characteristics of the complex network will be analyzed; then, based on the relationship network of the detainees, an information transmission model will be constructed to find out the key nodes in the network; Through various community discovery algorithms of complex networks, the detainees' relationship network is divided into communities, potential small groups are found, and the intergroup interaction patterns are analyzed; finally, based on the "key prisoners" and small groups analyzed above, an appropriate Evidence-based correction order and intensity to achieve better correction effect, and timely track and update correction activities to maximize the benefits of correction practice.

本发明基于在押人员关系网络指导循证矫治工作的方法，包括以下步骤:The present invention is based on the method for guiding evidence-based correction work based on the relationship network of detainees, comprising the following steps:

步骤1.建立在押人员关系网络，分析其统计特性；Step 1. Establish a network of detainees and analyze their statistical characteristics;

步骤2.基于在押人员关系网络建立信息传递模型，根据信息传递效率，找出“关键犯人”；Step 2. Establish an information transfer model based on the relationship network of detainees, and find "key prisoners" according to the efficiency of information transfer;

步骤3.对在押人员关系网络进行社团划分，分析团体行为；Step 3. Divide the network of detainees into groups and analyze group behavior;

步骤4.以“关键犯人”和小团体的分析为依据，设计适当的循证矫正。Step 4. Design appropriate evidence-based corrections based on analysis of "key offenders" and small groups.

步骤1.建立在押人员关系网络，分析其统计特性Step 1. Establish a network of detainees and analyze their statistical characteristics

1-1)收集可用于分析在押人员关系网络的数据；1-1) Collect data that can be used to analyze the network of detainees;

1-2)对数据进行初步分析整理，构建在押人员关系网络；1-2) Preliminary analysis and arrangement of the data to build a relationship network of detainees;

1-3)采用复杂网络的统计特征分析的方法，寻找网络中的关键节点；1-3) Use the method of statistical feature analysis of complex networks to find key nodes in the network;

步骤2.基于在押人员关系网络建立信息传递模型，根据信息传递效率，找出“关键犯人”Step 2. Establish an information transfer model based on the relationship network of detainees, and identify "key prisoners" based on the efficiency of information transfer

2-1)基于在押人员关系网络，构建信息传递模型；2-1) Based on the relationship network of detainees, build an information transfer model;

2-2)模拟信息在该网络上的传播过程，计算在押人员的信息传递效率；2-2) Simulate the dissemination process of information on the network, and calculate the information transmission efficiency of the detainees;

2-3)根据传递效率确定网络中的关键节点；2-3) Determine the key nodes in the network according to the transfer efficiency;

步骤3.对在押人员关系网络进行社团划分，分析团体行为Step 3. Divide the network of detainees into groups and analyze group behavior

3-1)基于在押人员关系网络，选择例如Fast greedy算法、随机游走算法、层次聚类算法等社团结构发现算法，对在押人员进行社团划分；3-1) Based on the relationship network of detainees, select community structure discovery algorithms such as Fast greedy algorithm, random walk algorithm, hierarchical clustering algorithm, etc., to divide detainees into communities;

3-2)计算每个算法的模块度，取模块度最高算法的社团划分结果作为最终结果；3-2) Calculate the modularity of each algorithm, and take the community division result of the algorithm with the highest modularity as the final result;

3-3)分析社团内部成员的社交关系和相处模式；3-3) Analyze the social relationships and get along patterns of members within the community;

步骤4.以“关键犯人”和小团体的分析为依据，设计适当的循证矫正Step 4. Design appropriate evidence-based corrections based on analysis of “key offenders” and small groups

4-1)基于上述的关键节点和社交关系，设计合理的矫正顺序和强度安排，跟踪了解矫正效果；4-1) Based on the above-mentioned key nodes and social relationships, design a reasonable correction order and intensity arrangement, and track and understand the correction effect;

4-2)及时根据新的分析情况做出调整，更新矫正顺序和强度安排。4-2) Make adjustments in time according to the new analysis situation, and update the correction sequence and intensity arrangement.

有益效果beneficial effect

1、相较于以往的在押人员相关研究，本方法采用复杂网络将在押人员之间的社交关系模型化，并通过复杂网络上统计特征分析、网络动力学等方法研究在押人员在群体中的地位和作用，为狱警的监控和管制提供新的方法和视角；1. Compared with previous researches on detainees, this method uses a complex network to model the social relationship between detainees, and studies the status of detainees in the group through statistical analysis of complex networks, network dynamics and other methods. and role, providing new methods and perspectives for the monitoring and control of prison guards;

2、相较于传统的信息传递模型，本方法提出了基于在押人员亲密度的信息传递模型，通过衡量在押人员的信息传递效率发现关键节点，指导狱警的监控和管制；2. Compared with the traditional information transmission model, this method proposes an information transmission model based on the intimacy of the detainees, and finds key nodes by measuring the information transmission efficiency of the detainees to guide the monitoring and control of prison guards;

3、采用复杂网络社团发现算法，可以对在押人员进行分组，便于研究其团体内部的相处模式，同时有利于对潜在的小团体进行监管。3. Using the complex network community discovery algorithm, the detainees can be grouped, which is convenient for studying the interpersonal pattern of their groups, and at the same time, it is beneficial to supervise potential small groups.

附图说明Description of drawings

图1基于在押人员关系网络指导循证矫治工作的方法流程图。Figure 1 is a flow chart of the method for guiding evidence-based corrections based on the detainee's relationship network.

具体实施方式Detailed ways

下面结合附图对本发明的技术方案进行详细说明：Below in conjunction with accompanying drawing, technical scheme of the present invention is described in detail:

本发明思路是收集在押人员相关数据，建立在押人员关系网络，分析其统计特性；然后建立信息传递模型，基于该在押人员关系网络模拟信息传递过程，根据信息传递效率，找出“关键犯人”；对在押人员关系网络进行社团划分，分析团体行为。The idea of the invention is to collect the relevant data of the detainees, establish the relationship network of the detainees, and analyze their statistical characteristics; then establish an information transmission model, simulate the information transmission process based on the relationship network of the detainees, and find "key prisoners" according to the efficiency of information transmission; Divide the network of detainees into groups and analyze group behavior.

本发明方法的基本流程如图1所示，具体包括以下步骤：The basic flow of the method of the present invention is shown in Figure 1, which specifically includes the following steps:

步骤1.建立在押人员关系网络，分析其统计特性；Step 1. Establish a network of detainees and analyze their statistical characteristics;

复杂网络的构建需要两个要素：节点和连边。每个在押人员即为节点，而连边则表示在押人员之间的社交关系。通过收集相关数据，确定节点和连边的相关属性，即可建立在押人员关系网络并进行统计分析。The construction of complex networks requires two elements: nodes and edges. Each detainee is a node, and the link represents the social relationship between the detainees. By collecting relevant data and determining the relevant attributes of nodes and edges, the relationship network of detainees can be established and statistical analysis can be performed.

1-1)在收集数据时，可以通过在押人员行为序列的相关性或者其他方式得到在押人员之间的联系——连边，甚至可以通过动态行为的重构的方式，最终建立在押人员关系网络。例如可以收集监控视频，分析视频中在押人员是否经常共处或者交流的紧密程度，确定其社交关系，可以收集犯人的社会信息、生理数据和心理数据等，对数据进行归一化处理，并根据犯人之间的相似度确定其潜在的社交关系等。1-1) When collecting data, the relationship between detainees can be obtained through the correlation of detainees' behavior sequences or other methods - linking, or even through dynamic behavior reconstruction, and finally establish a relationship network between detainees . For example, surveillance videos can be collected to analyze whether the detainees often coexist or communicate closely in the videos, determine their social relationships, collect social information, physiological data, and psychological data of prisoners, normalize the data, and analyze the data according to the prisoners. The similarity between them determines their potential social relationships, etc.

1-2)基于该在押人员关系网络，通过计算其每个节点的度值、集聚系数、节点介数中心性和节点接近中心性等指标，确定在押人员的关键指数。1-2) Based on the relationship network of detainees, determine the key index of detainees by calculating the degree value, agglomeration coefficient, node betweenness centrality and node proximity centrality of each node.

1-3)各个指标的具体含义如下：1-3) The specific meaning of each indicator is as follows:

度中心性：度中心性认为一个节点的邻居数目越多，影响力就越大，这是网络中刻画节点重要性最简单的指标。度中心性刻画的是节点的直接影响力，它认为一个节点的度越大，能直接影响的邻居就越多，也就越重要。值得注意的是，不同规模的网络中有相同度值的节点有不同的影响力，为了进行比较，定义节点v_i的归一化度中心性指标如公式(1)：Degree centrality: Degree centrality believes that the more neighbors a node has, the greater its influence, which is the simplest indicator to describe the importance of a node in the network. Degree centrality describes the direct influence of a node. It believes that the greater the degree of a node, the more neighbors it can directly influence, and the more important it is. It is worth noting that nodes with the same degree value in networks of different scales have different influences. For comparison, the normalized degree centrality index of node v_i is defined as formula (1):

其中，k_i为节点v_i的一阶邻居数，且k_i＝∑_ia_ij，a_ij即网络邻接矩阵A中第i行第j列元素，n为网络的节点数目，分母n-1为节点可能的最大度值。Among them,_ki is the number of first-order neighbors of node vi, and_ki = ∑_i a_ij , a_ij is the_i -th row and j-th column element in the network adjacency matrix A, n is the number of nodes in the network, and the denominator is n-1 is the maximum possible degree value of the node.

集聚系数：在图论和网络中,集聚系数是用来描述一个网络(图)中的顶点之间结集成团的程度的系数，具体来说，是一个点的邻接点之间相互连接的程度。一个人的集聚系数越大，代表他更有凝聚力，能够号召更多的人，关键指数较高。其中，无向图的集聚系数定义为：Agglomeration Coefficient: In graph theory and networks, the Agglomeration Coefficient is a coefficient used to describe the degree to which vertices in a network (graph) are grouped together, specifically, the degree to which adjacent points of a point are connected to each other . The larger a person's agglomeration coefficient is, the more cohesive he is, the more people he can call on, and the higher the key index. Among them, the clustering coefficient of the undirected graph is defined as:

其中，E为网络的边集，N_i为节点v_i的邻居集，e_jk为节点v_i的邻居节点之间的连边，k_i为节点v_i的一阶邻居数。Among them, E is the edge set of the network, N_i is the neighbor set of node v_i , e_jk is the connecting edge between the neighbor nodes of node v_i , and_ki is the number of first-order neighbors of node v_i .

节点接近中心性：在图论和网络中,节点接近中心性体现的是一个点与其他点的近邻程度。一个人的接近中心性越大，代表他与其他人的平均距离较小。接近中心性也可以理解为利用信息在网络中的平均传播时长来确定节点的重要性。平均来说，接近中心性最大的节点对于信息的流动具有最佳的观察视野。对于有n个节点的连通网络，可以计算任意一个节点v_i到网络中其他节点的平均最短距离：Node proximity centrality: In graph theory and networks, node proximity centrality reflects how close a point is to other points. The greater the proximity centrality of a person, the smaller the average distance between him and others. Proximity centrality can also be understood as using the average propagation time of information in the network to determine the importance of nodes. On average, nodes that are close to the most centrality have the best view of the flow of information. For a connected network with n nodes, the average shortest distance from any node v_i to other nodes in the network can be calculated:

d_i越小意味着节点v_i更接近网络中的其他节点，于是把d_i的倒数定义为节点v_i的接近中心性，即：The smaller d_i means the node v_i is closer to other nodes in the network, so the reciprocal of d_i is defined as the closeness centrality of the node v_i , namely:

如果节点v_i和v_j之间没有路径可达则定义d_ij＝∞，即1/d_ij＝0。接近中心性利用所有节点对之间的相对距离确定节点的中心性，在研究中应用非常广泛，但时间复杂度比较高。If there is no reachable path between nodes v_i and v_j , define d_ij =∞, that is, 1/d_ij =0. Proximity centrality uses the relative distance between all node pairs to determine the centrality of nodes, which is widely used in research, but the time complexity is relatively high.

节点介数中心性：通常提到的介数中心性一般指最短路径介数中心性，它认为网络中所有节点对的最短路径中(一般情况下一对节点之间存在多条最短路径)，经过一个节点的最短路径数越多，这个节点就越重要。介数中心性刻画了节点对网络中沿最短路径传输的网络流的控制力。节点v_i的介数定义为：Node betweenness centrality: The commonly mentioned betweenness centrality generally refers to the shortest path betweenness centrality, which considers that among the shortest paths of all pairs of nodes in the network (in general, there are multiple shortest paths between a pair of nodes), The greater the number of shortest paths through a node, the more important the node is. Betweenness centrality characterizes the power of nodes to control the network flow along the shortest path in the network. The betweenness of a node v_i is defined as:

其中，g_st为从节点v_s到v_t的所有最短路径的数目，

为从节点v_s到v_t的g_st条最短路径中经过v_i的最短路径的数目。where g_st is the number of all shortest paths from node v_s to v_t ,

is the number of shortest paths passing through vi in the_gst shortest paths from node v_s to v_t .

关键度：自定义指标，根据收集的数据定义每个节点的初始关键度，由于每个节点会受到网络中其他节点的影响，所以定义每个节点的关键度为自身初始关键度和其一阶邻居的关键度的加权和，即对于一个节点v_i，它的关键度为：Criticality: A custom indicator, which defines the initial criticality of each node according to the collected data. Since each node will be affected by other nodes in the network, the criticality of each node is defined as its own initial criticality and its first-order criticality. The weighted sum of the criticality of neighbors, that is, for a node v_i , its criticality is:

其中α,β为常数，且

r₀为节点i的初始关键度，N_i为节点v_i的一阶邻居集，k_i为节点v_i的一阶邻居数。where α and β are constants, and

r₀ is the initial criticality of node i,_Ni is the first-order neighbor set of node v_i , and_ki is the number of first-order neighbors of node v_i .

步骤2.基于在押人员关系网络建立信息传递模型，根据信息传递效率，找出“关键犯人”Step 2. Establish an information transfer model based on the relationship network of detainees, and identify "key prisoners" based on the efficiency of information transfer

网络上的信息传递通常采用经典的SI模型、SIS模型和SIR模型，也可以根据实际需要自己建立和改进模型。其中，在SI模型中，包含S人群和I人群，S人群已接收到信息，I人群未接收到信息，信息以一定的概率从S人群传递到I人群；在SIS模型中，包含SI模型的所有假设，并且S人群有一定的概率遗忘信息，重新转变成I人群；在SIR模型中，包含SI模型的所有假设，并且新增的R人群表示不再接收信息，I人群有一定概率转变为R人群，即该信息对其已经没有影响。除此之外，还可以建立基于亲密关系进行信息传递的模型，通常来说，信息在关系更亲密的人之间传递更容易，更高效，而在关系疏远的人之间传递更困难，更低效。The information transmission on the network usually adopts the classic SI model, SIS model and SIR model, and can also establish and improve the model according to actual needs. Among them, in the SI model, the S population and the I population are included, the S population has received information, and the I population has not received information, and the information is transmitted from the S population to the I population with a certain probability; in the SIS model, the SI model contains the information. All hypotheses, and the S population has a certain probability of forgetting information, and re-transforms into the I population; in the SIR model, all the assumptions of the SI model are included, and the newly added R population means that they no longer receive information, and the I population has a certain probability to transform into R population, i.e. the information has no effect on it. In addition to this, it is possible to model information transfer based on intimacy. Generally speaking, it is easier and more efficient to transfer information between people who are more closely related, while it is more difficult and more efficient to transfer information between people who are more distant. Inefficient.

2-1)在建立在押人员关系网络时，添加边的权重w，表示两个节点之间的亲密程度。每个节点有一定概率将信息传递给跟它有直接关系的邻居节点，并且传播的时间受到节点之间亲密关系的限制，节点之间越亲密，他们的距离就越小，信息传递就越快，反之，节点之间越疏离，他们呢的距离就越大，信息传递就越慢。2-1) When establishing the relationship network of detainees, the weight w of the edge is added to represent the degree of intimacy between the two nodes. Each node has a certain probability to transmit information to neighboring nodes that are directly related to it, and the propagation time is limited by the intimacy between nodes. The closer the nodes are, the smaller the distance between them and the faster the information transmission. , Conversely, the more distant the nodes are, the greater the distance between them, and the slower the information transmission.

2-2)根据建立的在押人员关系网络和设定的信息传递模型，模拟信息在在押人员之间的传递过程：初始化所需要的参数传播概率p和传播速度v，设定信息源，对于每一个已获取信息的节点i，找出这个节点的一阶邻居，对于每一个邻居j，给出一个0-1之间的随机数，如果随机数大于p，认为该邻居可以从这个节点接收到信息，再根据节点i和邻居j连边的权重w以及传播速度v，计算邻居j获取信息的时间点。若一个邻居在t₁和t₂时刻均从外界接收到信息，取较小的时间点作为该邻居获取信息的时间点。2-2) According to the established detainee relationship network and the set information transfer model, simulate the transfer process of information between detainees: initialize the required parameter propagation probability p and propagation speed v, set the information source, for each A node i that has acquired information, find the first-order neighbors of this node, and give a random number between 0-1 for each neighbor j. If the random number is greater than p, it is considered that the neighbor can receive from this node. information, and then calculate the time point at which neighbor j obtains information according to the weight w and propagation speed v of the connection between node i and neighbor j. If a neighbor receives information from the outside world at both times t₁ and t₂ , the smaller time point is taken as the time point when the neighbor obtains information.

2-3)重复2-2)的过程，直至整个在押人员关系网络中有一半的节点获取到信息或者信息传递时间步已达设定的最长时间限度T。此时计算每个节点将信息传递给了多少个其他节点，作为其信息传递效率。2-3) Repeat the process of 2-2) until half of the nodes in the entire detainee relationship network have obtained the information or the information transmission time step has reached the set maximum time limit T. At this time, calculate how many other nodes each node transmits information to, as its information transmission efficiency.

步骤3.对在押人员关系网络进行社团划分，分析团体行为Step 3. Divide the network of detainees into groups and analyze group behavior

3-1)选择例如Fast greedy算法、随机游走算法、层次聚类算法和标签传播算法等社团结构发现算法，对在押人员进行社团划分，计算其模块度Q，选取模块度最高的作为最终结果。模块度公式为：3-1) Select community structure discovery algorithms such as Fast greedy algorithm, random walk algorithm, hierarchical clustering algorithm and label propagation algorithm, divide the detainees into communities, calculate their modularity Q, and select the one with the highest modularity as the final result . The modularity formula is:

其中，C_i表示顶点v_i所属的社团，a_ij即网络邻接矩阵A中第i行第j列元素，m为网络的连边数，k_i为节点v_i的一阶邻居数，若节点v_i和v_j属于同一社团，则δ(C_i,C_j)＝1，若节点v_i和v_j属于不同社团，则δ(C_i,C_j)＝0。Among them, C_i represents the community to which the vertex v_i belongs, a_ij is the element in the ith row and the jth column of the network adjacency matrix A, m is the number of connected edges of the network,_{ki is the number of first-order neighbors of the node v i,} if the node v_i and v_j belong to the same community, then δ(C_i , C_j )=1, if nodes v_i and v_j belong to different communities, then δ(C_i , C_j )=0.

3-2)根据划分的社团，分析团体内在押人员的社交关系及其相处模式。3-2) According to the divided groups, analyze the social relations and the patterns of interpersonal relationships among the detainees in the group.

步骤4.以“关键犯人”和小团体的分析为依据，设计适当的循证矫正Step 4. Design appropriate evidence-based corrections based on analysis of “key offenders” and small groups

4-1)基于上述的“关键犯人”和小团体社交关系，优先对关键指数较高的在押人员进行矫正，适度增强矫正强度，并跟踪了解矫正效果。4-1) Based on the above-mentioned "key prisoners" and social relationships in small groups, priority should be given to the correction of detainees with higher key indexes, the correction intensity should be moderately increased, and the correction effect should be tracked.

4-2)根据矫正措施和效果，及时对新的在押人员关系网络进行进一步分析，调整更新矫正活动，以达到最优的矫正效果。4-2) According to the corrective measures and effects, further analyze the relationship network of new detainees in a timely manner, and adjust and update corrective activities to achieve the best corrective effect.

Claims (5)

1. The method for guiding evidence-based correction work based on the escorting personnel relationship network is characterized in that firstly, collected social relationship data of escorting personnel is preliminarily analyzed and sorted, the escorting personnel relationship network is established, and the statistical characteristic analysis of a complex network is carried out; then, constructing an information transfer model based on the escort relationship network, and finding out key nodes in the network; finally, carrying out community division on the escort relationship network through various community discovery algorithms of a complex network, and specifically comprising the following steps:

step 1, establishing an escort relationship network, and analyzing the statistical characteristics of the escort relationship network;

step 2, establishing an information transfer model based on the escort relationship network, and finding out a key prisoner according to the information transfer efficiency;

step 3, carrying out community division on the escort relationship network, and analyzing the community behaviors;

and 4, designing appropriate evidence-based correction based on analysis of key prisoners and small groups.

2. The method of claim 1, wherein step 1 comprises:

2-1) collecting data that can be used to analyze the network of escort relationships;

2-2) carrying out preliminary analysis and arrangement on the data to construct an escort personnel relationship network;

2-3) searching key nodes in the network by adopting a statistical characteristic analysis method of the complex network.

3. The method of claim 1, wherein the step 2 comprises:

3-1) constructing an information transfer model based on the escort relationship network;

3-2) simulating the information transmission process on the network, and calculating the information transmission efficiency of the escorting personnel;

3-3) determining key nodes in the network according to the transfer efficiency.

4. The method of claim 1, wherein step 3 comprises:

4-1) selecting a community structure discovery algorithm such as Fast greedy algorithm, random walk algorithm, hierarchical clustering algorithm and the like based on the escorting personnel relationship network to divide communities for the escorting personnel;

4-2) calculating the modularity of each algorithm, and taking a community division result of the algorithm with the highest modularity as a final result;

4-3) analyzing social relations and interaction patterns of members in the community.

5. The method of claim 1, wherein step 4 comprises:

5-1) designing a reasonable correction sequence and strength arrangement based on the key nodes and the social relation, and tracking and knowing the correction effect;

and 5-2) adjusting in time according to the new analysis condition, updating the correction sequence and the intensity arrangement, and realizing the benefit maximization of the correction practice.

Images