WO2022056955A1 - 一种基于不确定图的社区发现方法 - Google Patents

一种基于不确定图的社区发现方法 Download PDF

Info

Publication number
WO2022056955A1
WO2022056955A1 PCT/CN2020/118130 CN2020118130W WO2022056955A1 WO 2022056955 A1 WO2022056955 A1 WO 2022056955A1 CN 2020118130 W CN2020118130 W CN 2020118130W WO 2022056955 A1 WO2022056955 A1 WO 2022056955A1
Authority
WO
WIPO (PCT)
Prior art keywords
node
nodes
stable
uncertain
graph
Prior art date
Application number
PCT/CN2020/118130
Other languages
English (en)
French (fr)
Inventor
于东晓
李峰
王东标
罗琦
Original Assignee
山东大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 山东大学 filed Critical 山东大学
Priority to DE112020007405.6T priority Critical patent/DE112020007405T5/de
Publication of WO2022056955A1 publication Critical patent/WO2022056955A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90Details of database functions independent of the retrieved data types
    • G06F16/901Indexing; Data structures therefor; Storage structures
    • G06F16/9024Graphs; Linked lists
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90Details of database functions independent of the retrieved data types
    • G06F16/906Clustering; Classification

Definitions

  • the invention relates to the technical field of data mining, in particular to a community discovery method based on an uncertain graph.
  • a community is a dense subgraph composed of some closely connected nodes. The connections between nodes belonging to the same community are relatively close, while the connections between nodes in different communities are relatively sparse.
  • the community structure usually corresponds to the obvious clustering in the graph, so the community structure in the graph can be mined by clustering.
  • Structural clustering can be used to distinguish core nodes, hub nodes and outlier nodes. Among them, core nodes belonging to the same cluster are closely connected, while core nodes belonging to different clusters may have hub nodes or outliers node.
  • the structural clustering algorithm SCAN on the deterministic graph is developed from the density-based clustering algorithm DBSCAN.
  • DBSCAN is a commonly used clustering algorithm in multi-dimensional data. It uses the distance between data points as a measure of data similarity, and realizes the division of dense points and noisy points in the data according to two parameters ⁇ , ⁇ . Similar to the DBSCAN algorithm, the structural similarity is used in the SCAN algorithm to measure the degree of similarity between nodes. By the given two parameters ⁇ , ⁇ , the densely connected core nodes, hub nodes and outlier nodes in the graph can be obtained. Because the SCAN algorithm is sensitive to parameters and has high computational complexity, some subsequent structural clustering algorithms such as PSCAN, SCAN++, etc. have improved the SCAN algorithm from different aspects, so that it can be applied to large-scale graph data. But because these algorithms cannot be directly applied to structural clustering on uncertain graphs, these algorithms cannot be used for community discovery on uncertain graphs either.
  • the concept of structural clustering on deterministic graphs is extended to uncertain graphs.
  • the concept of structural similarity is extended from deterministic graphs to uncertain graphs, and a dynamic programming algorithm is designed to calculate the reliability measure between nodes, and finally Structure clustering on uncertain graph is realized by modifying the framework of SCAN algorithm. After the clusters on the uncertainty graph are obtained, the clustering results can be output as the community structure.
  • a limitation of this method is that it cannot calculate the probability that a node belongs to a core node, which may lead to a situation: although a node is a reliable core node, it has a high probability of being a core node in all possible worlds. Low.
  • the present invention provides a community discovery method based on an uncertain graph, which solves the problem of insufficient reliability of the core nodes in the prior art by constraining the reliability of the core nodes in the uncertain graph.
  • a community discovery method based on uncertainty graph including the following steps:
  • Step 1 Collect uncertain data in the actual network and construct an uncertain graph, which is stored in the form of an adjacency linked list;
  • Step 2 use the stable structure clustering algorithm to mine stable core nodes, pivot nodes and outlier nodes in the uncertain graph, and construct stable structure clusters from the stable core nodes;
  • the third step is to construct communities from the stable structure clustering, that is, the community structure in the uncertain graph.
  • uncertain data is collected from the actual network, and after measuring these uncertain data, if the probability of the edge between two nodes u, v is not 0, then in the adjacency list A triple (u, v, p) is stored in , where p represents the probability of edge existence.
  • the similarity threshold ⁇ between nodes is used to constrain the structural similarity between nodes;
  • the probability threshold ⁇ the range of this value is 0 ⁇ 1, used to judge whether two neighbor nodes can become reliable neighbors, whether the edge where the two neighbor nodes are located can become a reliable edge, and whether a node can become a stable core node;
  • the threshold of the number of reliable neighbors ⁇ the range of this value is ⁇ 2, which is used to constrain the number of reliable neighbors of a node.
  • step (2) according to the probability structure similarity formula Calculate the probabilistic structural similarity of each edge, where represents the uncertainty graph, means that G is possible worlds, for G in the probability of occurrence in ;
  • N[u] is the number of neighbors of node u
  • N[u] ⁇ N[v] is the intersection of the neighbors of node u
  • v N[u] ⁇ N[v] is the union of neighbors of node u, v
  • I( ⁇ (e) ⁇ ) is an indicator function whose value is 1 if ⁇ (e) ⁇ , and 0 otherwise;
  • step (3) for each node, the adjacent edges of the node are sorted in descending order of probability structure similarity. Multiply their probability structure similarity, if the product is greater than or equal to ⁇ , the node belongs to the stable core node, and marks the node as unvisited, otherwise, the node belongs to the unstable core node.
  • step (4) for an unvisited stable core node, starting from the stable core node, visit all stable core nodes that can be reached by the node through reliable edges and mark them as visited.
  • the core nodes belong to the same cluster; if there are stable core nodes that have not been visited, repeat the process until all stable core nodes are marked as visited;
  • the unstable core node For each unstable core node, if the edge between the node and a stable core node is a reliable edge, the unstable core node is assigned to the cluster where the connected stable core node is located.
  • step (5) for each unstable core node that does not belong to any cluster, if it can connect two or more clusters, the node belongs to the hub node, otherwise the node belongs to the outlier. node.
  • step 3 the stable structure cluster found by the stable structure clustering algorithm is used as the community structure output.
  • the present invention also considers the reliability of the edges in the uncertain graph and the reliability of the core nodes.
  • the stable core nodes in the community and the reliable edges connecting them can meet certain reliability guarantees .
  • the present invention makes full use of the properties of the uncertain graph, designs an efficient stable structure clustering algorithm to calculate the stable core nodes in the uncertain graph, and mines the community structure on this basis, with low computational complexity and can be applied into large-scale graph data.
  • FIG. 1 is a schematic overall flowchart of a method for community discovery based on an uncertain graph disclosed in an embodiment of the present invention
  • Figure 2 is a schematic diagram of the main flow of the stable structure discovery stage
  • FIG. 3 is a schematic diagram of a specific flow of the stable structure discovery stage.
  • the present invention provides a community discovery method based on uncertainty graph, as shown in FIG. 1 , and the specific embodiment is as follows:
  • Uncertain data can be generated in many ways, such as measurement errors, errors in machine learning models, etc.
  • users represent nodes, and the relationship between users represents edges.
  • a probability value is used to indicate the strength of the relationship between users. The higher the probability, the closer the relationship between two users is.
  • proteins represent nodes, and edges between nodes represent interactions between proteins.
  • the error generated during measurement can be represented by probability. The greater the probability, the more credible the measurement result is.
  • Set the similarity threshold ⁇ between nodes is used to constrain the structural similarity between nodes; set the probability threshold ⁇ , the range of this value is 0 ⁇ 1 , which is used to judge whether two neighbor nodes can become reliable neighbors (the edges where these two neighbors are located are reliable edges) and whether a node can become a stable core node; set the threshold value of the number of reliable neighbors ⁇ , and the range of this value is ⁇ ⁇ 2, used to constrain the number of reliable neighbors of a node.
  • I( ⁇ (e) ⁇ ⁇ ) is an indicator function whose value is 1 if ⁇ (e) ⁇ ⁇ , and 0 otherwise. For each edge e, if Pr[e, ⁇ ] ⁇ , then the edge is a reliable edge.
  • the node is a stable core node, and the node is marked as unvisited, otherwise, the node is unstable core node.
  • the stable structure clusters found in step 2 are output as community structures.

Abstract

本发明公开了一种基于不确定图的社区发现方法,包括如下步骤:步骤一,采集实际网络中的不确定数据并构建不确定图,以邻接链表的形式进行存储;步骤二,使用稳定结构聚类算法挖掘不确定图中的稳定核心节点,枢纽节点和离群节点,并从稳定核心节点中构建稳定结构聚类;步骤三,从稳定结构聚类中构建社区,即不确定图中的社区结构。本发明所公开的方法同时考虑了不确定图中边的可靠性和核心节点的可靠性,在最终得到的社区中,社区中的稳定核心节点以及连接它们的可靠边均能够满足一定的可靠性保证;并且本发明采用高效的稳定结构聚类算法来计算不确定图中的稳定核心节点,并在此基础上挖掘社区结构,计算复杂度低,能够应用到大规模的图数据中。

Description

一种基于不确定图的社区发现方法 技术领域
本发明涉及数据挖掘技术领域,特别涉及一种基于不确定图的社区发现方法。
背景技术
作为一种能够表示关联关系的数据范式,随着社会和科学技术的发展,越来越多的数据能够通过图模型来进行建模、表示和组织。由于技术限制和实际需要,如测量误差,方法中带有的误差等,数据之间的关系通常带有不确定性。这种不确定性可以使用一个概率值p,p∈(0,1]来表示。这种带有不确定性的图模型可以使用不确定图来进行建模和分析。
考虑一个无权无向的不确定图
Figure PCTCN2020118130-appb-000001
其中V是节点集合,E是边集合。对于E中的每一条边e,存在概率P e和它对应代表该边在不确定图中存在的概率。我们使用可能世界模型来解释和分析不确定图。在可能世界模型中,每一个可能世界都是节点数为V的确定图。我们使用
Figure PCTCN2020118130-appb-000002
来表示G是
Figure PCTCN2020118130-appb-000003
的一个可能世界,其中G=(V,E G),并且
Figure PCTCN2020118130-appb-000004
社区是一些紧密连接的节点构成的稠密子图,属于同一个社区的节点之间的联系相对紧密,而在不同社区的节点之间的联系比较稀疏。在图模型中,社区结构通常也对应着图中明显的聚类,所以可以使用聚类的方式挖掘图中的社区结构。
结构聚类可以用来区分核心节点,枢纽节点和离群节点,其中,属于同一个聚类的核心节点之间连接比较紧密,而属于不同聚类的核心节点之间会存在枢纽节点或离群节点。
确定图上的结构聚类算法SCAN是由基于密度的聚类算法DBSCAN发展而来。DBSCAN是一个在多维数据中常用的聚类算法,它将数据点之间的距离作为数据的相似性度量,并根据两个参数∈,μ实现了对数据中密集点和噪声点的划分。和DBSCAN算法类似,在SCAN算法中使用了结构相似性来衡量节点之间的相似程度,通过给定的两个参数∈,μ可以得到图中密集连接的核心节点,枢纽节点和离群节点。由于SCAN算法对参数较为敏感并且计算复杂度较高,后续的一些结构聚类算法如PSCAN,SCAN++等分别从不同的方面对SCAN算法进行了改进,使它能够应用到大规模的图数据中。但是因为这些算法不能直接的应用于不确定图上的结构聚类,所以这些算法也不能够用于不确定图上的社区发现。
在USCAN算法中,将确定图上结构聚类的概念扩展到不确定图中。通过在不确定图中定义概率结构相似度和可靠性度量,将结构相似性的概念从确定图扩展到不确定图,并设计 了一个动态规划算法来对节点间的可靠性度量进行计算,最后通过修改SCAN算法的框架实现了不确定图上的结构聚类。在得到了不确定图上的聚类之后,便可以将聚类结果作为社区结构输出。这种方法存在的一个局限是无法计算一个节点属于核心节点的概率,这就可能导致一种情况:虽然一个节点是一个可靠的核心节点,但是它在所有的可能世界中是核心节点的概率很低。
发明内容
为解决上述技术问题,本发明提供了一种基于不确定图的社区发现方法,通过对不确定图中核心节点的可靠性进行约束,解决了现有技术中核心节点存在可靠性不足的问题。
为达到上述目的,本发明的技术方案如下:
一种基于不确定图的社区发现方法,包括如下步骤:
步骤一,采集实际网络中的不确定数据并构建不确定图,以邻接链表的形式进行存储;
步骤二,使用稳定结构聚类算法挖掘不确定图中的稳定核心节点,枢纽节点和离群节点,并从稳定核心节点中构建稳定结构聚类;
步骤三,从稳定结构聚类中构建社区,即不确定图中的社区结构。
上述方案中,所述步骤一中,从实际网络中采集不确定数据,在对这些不确定数据进行测量后,如果两个节点u,v之间边上的概率不为0,则在邻接链表中保存一个三元组(u,v,p),p表示边存在的概率。
上述方案中,所述步骤二的具体方法如下:
(1)初始化稳定结构聚类算法中的参数,包括节点之间的相似度阈值∈,概率阈值η,可靠邻居的个数阈值μ;
(2)计算不确定图中每条边的概率结构相似度和不确定图中的可靠边;
(3)计算不确定图中的稳定核心节点;
(4)计算稳定结构聚类;
(5)计算枢纽节点和离群节点;
(6)输出聚类结果。
上述方案中,步骤(1)中,节点之间的相似度阈值∈,该值的范围是0<∈≤1,用来约束节点之间的结构相似性;概率阈值η,该值的范围是0<η≤1,用来判断两个邻居节点能否成为可靠邻居,这两个邻居节点所在的边能否成为可靠边,和一个节点能否成为稳定核心 节点;可靠邻居的个数阈值μ,该值的范围是μ≥2,用来约束一个节点可靠邻居的个数。
上述方案中,步骤(2)中,根据概率结构相似度公式
Figure PCTCN2020118130-appb-000005
计算每条边的概率结构相似度,其中
Figure PCTCN2020118130-appb-000006
代表不确定图,
Figure PCTCN2020118130-appb-000007
表示G是
Figure PCTCN2020118130-appb-000008
的可能世界,
Figure PCTCN2020118130-appb-000009
代表G在
Figure PCTCN2020118130-appb-000010
中的出现概率;
对于边e=(u,v)来说,
Figure PCTCN2020118130-appb-000011
N[u]表示为节点u的邻居个数,N[u]∩N[v]是节点u,v邻居的交集,N[u]∪N[v]是节点u,v邻居的并集;I(σ(e)≥∈)是一个指示函数,如果σ(e)≥∈,它的值为1,否则为0;
如果Pr[e,∈]≥η,那么该边是可靠边。
上述方案中,步骤(3)中,对于每一个节点,将节点的邻接边按照概率结构相似度降序排序,对于每一个邻居个数大于等于μ的节点,选择排序后的前μ条邻接边并将它们的概率结构相似度相乘,如果乘积大于等于η,则该节点属于稳定核心节点,并标记节点为未访问,否则,该节点属于非稳定核心节点。
上述方案中,步骤(4)中,对于一个未被访问过的稳定核心节点,从该稳定核心节点出发,访问能够被该节点通过可靠边到达的所有稳定核心节点并标记为已访问,这些稳定核心节点属于同一个聚类;如果存在未被访问过的稳定核心节点,重复该过程,直到所有的稳定核心节点都被标记为已访问;
对于每一个非稳定核心节点,如果该节点和一个稳定核心节点之间的边为可靠边,则将该非稳定核心节点分配到所连接的稳定核心节点所在的聚类中。
上述方案中,步骤(5)中,对于每一个不属于任何聚类的非稳定核心节点,如果它能够连接两个及两个以上的聚类,该节点属于枢纽节点,否则该节点属于离群节点。
上述方案中,步骤三中,将稳定结构聚类算法发现的稳定结构聚类作为社区结构输出
通过上述技术方案,本发明提供的一种基于不确定图的社区发现方法具有如下有益效果:
(1)本发明同时考虑了不确定图中边的可靠性和核心节点的可靠性,在最终得到的社区中,社区中的稳定核心节点以及连接它们的可靠边均能够满足一定的可靠性保证。
(2)本发明充分利用不确定图的性质,设计了一个高效的稳定结构聚类算法来计算不确定图中的稳定核心节点,并在此基础上挖掘社区结构,计算复杂度低,能够应用到大规模的图数据中。
附图说明
为了更清楚地说明本发明实施例或现有技术中的技术方案,下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍。
图1为本发明实施例所公开的一种基于不确定图的社区发现方法整体流程示意图;
图2为稳定结构发现阶段的主要流程示意图;
图3为稳定结构发现阶段的具体流程示意图。
具体实施方式
下面将结合本发明实施例中的附图,对本发明实施例中的技术方案进行清楚、完整地描述。
本发明提供了一种基于不确定图的社区发现方法,如图1所示,具体实施例如下:
1.网络构建
不确定数据的产生可能由多种方式,如测量误差,机器学习模型中的误差等。比如在社交网络中,用户代表节点,用户之间的关系代表边,使用一个概率值表示用户之间关系的强弱,概率越大,两个用户之间的关系越紧密。在蛋白质交互网络中,蛋白质代表节点,节点之间的边代表蛋白质之间存在相互作用,在测量时产生的误差可以使用概率来表示,概率越大,代表测量的结果越可信。在对这些不确定数据进行测量后,如果两个节点(例如节点u,v)之间边上的概率不为0,则在邻接链表中保存一个三元组(u,v,p),p表示边存在的概率。
2.稳定结构发现,如图2和图3所示,具体过程如下:
(1)设置节点之间的相似度阈值∈,该值的范围是0<∈≤1,用来约束节点之间的结构相似性;设置概率阈值η,该值的范围是0<η≤1,用来判断两个邻居节点能否成为可靠邻居(这两个邻居所在的边为可靠边)和一个节点能否成为稳定核心节点;设置可靠邻居的个数阈值μ,该值的范围是μ≥2,用来约束一个节点可靠邻居的个数。
(2)根据概率结构相似度公式
Figure PCTCN2020118130-appb-000012
计算每条边的概率结构相似度,其中
Figure PCTCN2020118130-appb-000013
代表不确定图,
Figure PCTCN2020118130-appb-000014
表示G是
Figure PCTCN2020118130-appb-000015
的可能世界,
Figure PCTCN2020118130-appb-000016
代表G在
Figure PCTCN2020118130-appb-000017
中的出现概率。对于边e=(u,v)来说,
Figure PCTCN2020118130-appb-000018
N[u]表示为节点u的邻居个数,N[u]∩N[v]是节点u,v邻居的交集,N[u]∪N[v]是节点u,v邻居的并集。
I(σ(e)≥∈)是一个指示函数,如果σ(e)≥∈,它的值为1,否则为0。对于每一条边e,如果Pr[e,∈]≥η,那么该边是可靠边。
(3)选择一个尚未处理的节点,将节点的邻接边按照概率结构相似度降序排序。
(4)选择排序后的前μ条邻接边并将它们的概率结构相似度相乘,如果乘积大于等于η,该节点属于稳定核心节点,并标记节点为未访问,否则,该节点属于非稳定核心节点。
(5)如果存在未处理的节点,重复步骤(4),否则执行(6)。
(6)选择一个未被访问过的稳定核心节点,从该稳定核心节点出发,访问能够被该节点通过可靠边到达的所有稳定核心节点并标记为已访问,这些稳定核心节点属于同一个聚类。
(7)如果存在未被访问过的稳定核心节点,重复(6),否则执行(8)。
(8)对于每一个非稳定核心节点,如果该节点和一个稳定核心节点之间的边为可靠边,则将该非稳定核心节点分配到所连接的稳定核心节点所在的聚类中。
(9)对于每一个不属于任何聚类的非稳定核心节点,如果它能够连接两个及两个以上的聚类,该节点属于枢纽节点,否则该节点属于离群节点。
3.社区构建
将步骤2中发现的稳定结构聚类作为社区结构输出。
对所公开的实施例的上述说明,使本领域专业技术人员能够实现或使用本发明。对这些实施例的多种修改对本领域的专业技术人员来说将是显而易见的,本文中所定义的一般原理可以在不脱离本发明的精神或范围的情况下,在其它实施例中实现。因此,本发明将不会被限制于本文所示的这些实施例,而是要符合与本文所公开的原理和新颖特点相一致的最宽的范围。

Claims (9)

  1. 一种基于不确定图的社区发现方法,其特征在于,包括如下步骤:
    步骤一,采集实际网络中的不确定数据并构建不确定图,以邻接链表的形式进行存储;
    步骤二,使用稳定结构聚类算法挖掘不确定图中的稳定核心节点,枢纽节点和离群节点,并从稳定核心节点中构建稳定结构聚类;
    步骤三,从稳定结构聚类中构建社区,即不确定图中的社区结构。
  2. 根据权利要求1所述的一种基于不确定图的社区发现方法,其特征在于,所述步骤一中,从实际网络中采集不确定数据,在对这些不确定数据进行测量后,如果两个节点u,v之间边上的概率不为0,则在邻接链表中保存一个三元组(u,v,p),p表示边存在的概率。
  3. 根据权利要求1所述的一种基于不确定图的社区发现方法,其特征在于,所述步骤二的具体方法如下:
    (1)初始化稳定结构聚类算法中的参数,包括节点之间的相似度阈值ε,概率阈值η,可靠邻居的个数阈值μ;
    (2)计算不确定图中每条边的概率结构相似度和不确定图中的可靠边;
    (3)计算不确定图中的稳定核心节点;
    (4)计算稳定结构聚类;
    (5)计算枢纽节点和离群节点;
    (6)输出聚类结果。
  4. 根据权利要求3所述的一种基于不确定图的社区发现方法,其特征在于,步骤(1)中,节点之间的相似度阈值ε,该值的范围是0<ε≤1,用来约束节点之间的结构相似性;概率阈值η,该值的范围是0<η≤1,用来判断两个邻居节点能否成为可靠邻居,这两个邻居节点所在的边能否成为可靠边,和一个节点能否成为稳定核心节点;可靠邻居的个数阈值μ,该值的范围是μ≥2,用来约束一个节点可靠邻居的个数。
  5. 根据权利要求3所述的一种基于不确定图的社区发现方法,其特征在于,步骤(2)中,根据概率结构相似度公式
    Figure PCTCN2020118130-appb-100001
    计算每条边的概率结构相似度,其中
    Figure PCTCN2020118130-appb-100002
    代表不确定图,
    Figure PCTCN2020118130-appb-100003
    表示G是
    Figure PCTCN2020118130-appb-100004
    的可能世界,
    Figure PCTCN2020118130-appb-100005
    代表G在
    Figure PCTCN2020118130-appb-100006
    中的出现概率;
    对于边e=(u,v)来说,
    Figure PCTCN2020118130-appb-100007
    N[u]表示为节点u的邻居个数,N[u]∩N[v]是节点u,v邻居的交集,N[u]∪N[v]是节点u,v邻居的并集;I(σ(e)≥ε)是一个指示函数,如果 σ(e)≥ε,它的值为1,否则为0;
    如果Pr[e,ε]≥η,那么该边是可靠边。
  6. 根据权利要求5所述的一种基于不确定图的社区发现方法,其特征在于,步骤(3)中,对于每一个节点,将节点的邻接边按照概率结构相似度降序排序,对于每一个邻居个数大于等于μ的节点,选择排序后的前μ条邻接边并将它们的概率结构相似度相乘,如果乘积大于等于η,则该节点属于稳定核心节点,并标记节点为未访问,否则,该节点属于非稳定核心节点。
  7. 根据权利要求6所述的一种基于不确定图的社区发现方法,其特征在于,步骤(4)中,对于一个未被访问过的稳定核心节点,从该稳定核心节点出发,访问能够被该节点通过可靠边到达的所有稳定核心节点并标记为已访问,这些稳定核心节点属于同一个聚类;如果存在未被访问过的稳定核心节点,重复该过程,直到所有的稳定核心节点都被标记为已访问;
    对于每一个非稳定核心节点,如果该节点和一个稳定核心节点之间的边为可靠边,则将该非稳定核心节点分配到所连接的稳定核心节点所在的聚类中。
  8. 根据权利要求7所述的一种基于不确定图的社区发现方法,其特征在于,步骤(5)中,对于每一个不属于任何聚类的非稳定核心节点,如果它能够连接两个及两个以上的聚类,该节点属于枢纽节点,否则该节点属于离群节点。
  9. 根据权利要求1所述的一种基于不确定图的社区发现方法,其特征在于,步骤三中,将稳定结构聚类算法发现的稳定结构聚类作为社区结构输出。
PCT/CN2020/118130 2020-09-16 2020-09-27 一种基于不确定图的社区发现方法 WO2022056955A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
DE112020007405.6T DE112020007405T5 (de) 2020-09-16 2020-09-27 Verfahren zum Auffinden einer Gemeinschaft basierend auf einem unsicheren Graphen

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010971870.9 2020-09-16
CN202010971870.9A CN112182306B (zh) 2020-09-16 2020-09-16 一种基于不确定图的社区发现方法

Publications (1)

Publication Number Publication Date
WO2022056955A1 true WO2022056955A1 (zh) 2022-03-24

Family

ID=73921531

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/118130 WO2022056955A1 (zh) 2020-09-16 2020-09-27 一种基于不确定图的社区发现方法

Country Status (3)

Country Link
CN (1) CN112182306B (zh)
DE (1) DE112020007405T5 (zh)
WO (1) WO2022056955A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112817963A (zh) * 2019-10-30 2021-05-18 华东师范大学 一种多维网络上的社群核分解方法及系统
CN117808616A (zh) * 2024-02-28 2024-04-02 中国传媒大学 一种基于图嵌入和节点亲密度的社区发现方法及系统

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115795527B (zh) * 2022-10-20 2023-08-18 福建师范大学 一种基于节点相似性的不确定图用户隐私保护方法

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140149430A1 (en) * 2012-11-28 2014-05-29 Korea Advanced Institute Of Science And Technology Method of detecting overlapping community in network
CN104573730A (zh) * 2015-01-29 2015-04-29 中国科学院深圳先进技术研究院 一种基于决定路径权重的不确定图分割方法及系统
CN106980795A (zh) * 2017-02-28 2017-07-25 广西师范大学 社会网络数据隐私保护方法

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140149430A1 (en) * 2012-11-28 2014-05-29 Korea Advanced Institute Of Science And Technology Method of detecting overlapping community in network
CN104573730A (zh) * 2015-01-29 2015-04-29 中国科学院深圳先进技术研究院 一种基于决定路径权重的不确定图分割方法及系统
CN106980795A (zh) * 2017-02-28 2017-07-25 广西师范大学 社会网络数据隐私保护方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
"Master's Thesis of Shenzhen University", 15 July 2019, SHENZHEN UNIVERSITY, CN, article QIU YUXUAN: "Research and Implementation of Structural Clustering on Probabilistic Graphs", XP055913809 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112817963A (zh) * 2019-10-30 2021-05-18 华东师范大学 一种多维网络上的社群核分解方法及系统
CN112817963B (zh) * 2019-10-30 2022-10-25 华东师范大学 一种多维网络上的社群核分解方法及系统
CN117808616A (zh) * 2024-02-28 2024-04-02 中国传媒大学 一种基于图嵌入和节点亲密度的社区发现方法及系统

Also Published As

Publication number Publication date
CN112182306A (zh) 2021-01-05
CN112182306B (zh) 2022-03-18
DE112020007405T5 (de) 2023-04-27

Similar Documents

Publication Publication Date Title
WO2022056955A1 (zh) 一种基于不确定图的社区发现方法
Chen et al. Personalized federated learning with graph
Lu Improved K-means clustering algorithm for big data mining under Hadoop parallel framework
Pons et al. Post-processing hierarchical community structures: Quality improvements and multi-scale view
Cheng et al. A local information based multi-objective evolutionary algorithm for community detection in complex networks
CN109086291B (zh) 一种基于MapReduce的并行异常检测方法及系统
US8228326B2 (en) Three dimensional surface and solid query processing
CN107133176A (zh) 一种基于半监督聚类数据筛选的跨项目缺陷预测方法
Liu et al. Dynamic local search based immune automatic clustering algorithm and its applications
CN109545275B (zh) 基于模糊谱聚类的不确定ppi网络功能模块挖掘方法
CN105678626B (zh) 重叠社区挖掘方法与装置
Luo et al. Distributed core decomposition in probabilistic graphs
CN106682514B (zh) 基于子图挖掘的系统调用序列特征模式集生成方法
Han et al. Generating uncertain networks based on historical network snapshots
Liu et al. Social group query based on multi-fuzzy-constrained strong simulation
CN114722947A (zh) 一种基于近邻搜索分簇的电力调度监控数据异常检测方法
CN113343123A (zh) 一种生成对抗多关系图网络的训练方法和检测方法
CN114329867B (zh) 一种基于模体的无标度网络鲁棒性度量方法
CN106778872B (zh) 基于密度的连通图的聚类方法与装置
CN107886107B (zh) 一种大数据的融合方法、系统和装置
CN109039722B (zh) 基于共同邻节点资源分配和朴素贝叶斯的链路预测方法
CN109086373B (zh) 一种构建公平的链接预测评估系统的方法
Chen et al. Robust truth discovery scheme based on mean shift clustering algorithm
Janeja et al. Spatial outlier detection in heterogeneous neighborhoods
Liu et al. Social Network Community-Discovery Algorithm Based on a Balance Factor

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20953813

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 20953813

Country of ref document: EP

Kind code of ref document: A1