CN112989272B - Community discovery algorithm based on local path - Google Patents
Community discovery algorithm based on local path Download PDFInfo
- Publication number
- CN112989272B CN112989272B CN202011623050.7A CN202011623050A CN112989272B CN 112989272 B CN112989272 B CN 112989272B CN 202011623050 A CN202011623050 A CN 202011623050A CN 112989272 B CN112989272 B CN 112989272B
- Authority
- CN
- China
- Prior art keywords
- node
- community
- local
- modularity
- nodes
- Prior art date
- Legal status (The legal status 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 status listed.)
- Active
Links
- 238000000034 method Methods 0.000 claims abstract description 12
- 238000012216 screening Methods 0.000 claims abstract description 6
- 238000012545 processing Methods 0.000 claims abstract description 5
- 239000011159 matrix material Substances 0.000 claims description 20
- 238000007689 inspection Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 8
- 230000007547 defect Effects 0.000 abstract description 6
- 238000004364 calculation method Methods 0.000 description 7
- 238000011160 research Methods 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000012217 deletion Methods 0.000 description 2
- 230000037430 deletion Effects 0.000 description 2
- 230000008030 elimination Effects 0.000 description 2
- 238000003379 elimination reaction Methods 0.000 description 2
- 230000008520 organization Effects 0.000 description 2
- 239000013589 supplement Substances 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000004900 laundering Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Databases & Information Systems (AREA)
- Algebra (AREA)
- Computing Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention belongs to the technical field of big data processing, and particularly relates to a community discovery algorithm based on a local path, which aims to solve the defects caused by the fact that the traditional community discovery algorithm starts from the global, and is characterized in that the method is increased in a gradually spreading way from local designated nodes, particularly, the community division on a larger network structure is very practical and effective, the discovery process is efficient and equivalent to the global community discovery in effect, relevant communities are quickly discovered through input nodes, the community screening is performed by adding time weak association in community judgment, the community misjudgment rate can be effectively reduced, the local community effect is improved, the relevance in communities is strong but non-community members can be reasonably pruned, and the community discovery accuracy is enhanced.
Description
Technical Field
The invention belongs to the technical field of big data processing, and particularly relates to a community discovery algorithm based on a local path.
Background
Many practical networks have been found to have a community structure, i.e. the entire network is made up of several communities, with relatively sparse connections between communities and relatively dense connections within communities. The community discovery is to analyze the modularized community structure from the complex network by utilizing the information stored in the topological structure of the graph, and the deep research of the problem is helpful to research the modules, functions and evolution of the whole network in a divide-and-conquer mode, so that the organization principle, topological structure and dynamics characteristics of the complex system can be more accurately understood, and the method has very important significance.
Banks master more data resources, and the nodes are hundreds of millions, in the whole network, if communities are found from the global angle, the calculation amount requirement is large, the communities are not well controlled in modularity, the obtained communities are more in modularity setting, the modules are low, and the money laundering communities possibly contain more normal users with relations.
The traditional community discovery algorithm generally performs community division on a global network, and along with the expansion of the network scale, the defects of the traditional algorithm appear, and the main performances are as follows: the model has low performance and low efficiency, and can not meet the demands of people, so that the research on a community discovery algorithm based on local paths is necessary.
Disclosure of Invention
Aiming at the defects and problems of the existing equipment, the invention provides a community discovery algorithm based on a local path, which effectively solves the problems of low model performance and low efficiency existing in the existing equipment.
The invention solves the technical problems by adopting the scheme that: a local path-based community discovery algorithm comprising the steps of:
step 1, data acquisition
Acquiring a designated node from a database, and acquiring an associated node which has direct correlation or primary indirect correlation with the designated node;
step 2: data preprocessing
The method comprises the steps of obtaining an original relation between a designated node and an associated node, and performing deduplication on the original relation based on weak deduplication logic of time postductility and data equality to obtain an effective relation between the designated node and the associated node; then designating the number of times of the current and the total value of all the current and the total value in the effective relation determination time period as an edge index;
step 3, calculating local similarity coefficients
If all the times of the node in the determined time period exceed the preset times y and the total value exceeds the preset value x, determining that the relationship between the node and the associated node is a, otherwise, b is b, and a+b=1; then obtaining the adjacency matrix A of the node network according to the values of a and b ij ；
Calculating local similarity coefficients according to the third-order adjacency matrix:
wherein lambda is the maximum characteristic value, A ij Is an adjacency matrix, S ij Is a local similarity coefficient;
step 4, comparing the local similarity coefficient with a threshold value
Comparing the local similarity coefficient obtained in the step 3 with a set threshold, if the local similarity coefficient is larger than the set threshold, adding the node network into a community, otherwise, not adding; performing community expansion through local coefficient judgment until all nodes are divided;
step 5, obtaining the rationality of the community by adopting local modularity inspection
Defining node set of community as V, V * V is the adjacent node of V * Is the adjacent matrix of (a)
Locating local modularity
Wherein delta (c) i ，c j ) If i, j are both in V, then 1, otherwise 0; m is m * Identifying a number of edges within the adjacency matrix;
if the modularity is larger than a preset modularity threshold, the community rationality is indicated to meet the requirements, otherwise, the next step is carried out;
step 6, processing communities with unqualified rationality
Sorting according to the degree of nodes in communities, performing time weak association judgment on the node with the smallest degree, if the weak association judgment meets the requirement, keeping the node in communities, removing the node from the sorting, and performing weak association judgment on the node with the smallest degree again; if the weak association judgment is not satisfactory, the node is removed from the community, and the modularity is recalculated and compared according to the mode in the step 5 until the modularity is greater than the modularity threshold.
Further, in step 1, the designated node is obtained through big data screening.
Further, in step 2, the determined period of time is one month or one year.
Further, in step 3, a is 1 and b is 0.
The invention has the beneficial effects that: the invention performs duplication elimination on the original relationship based on weak duplication elimination logic of time delay and data equality, improves the accuracy of data, designates the times of arrival and total data values as edge indexes, then performs local similarity coefficient calculation by using a third-order adjacency matrix, compares the local similarity coefficient with a set threshold value, and uses the local similarity coefficient as a basis of whether nodes are added into communities or not, without acquiring the whole network information, uses known characteristic nodes as break ports, and calculates the multi-dimensional index proportion of the nodes adjacent to the nodes one by one as a classification basis.
Meanwhile, the algorithm community algorithm can also judge according to the newly added nodes, and automatically optimize the whole community in a modular degree, so that the quality of members in the community is ensured.
Therefore, the invention provides a brand-new local community discovery algorithm, which aims to solve the defects brought by the traditional community discovery algorithm from the global point, adopts the gradual spreading type increase from the local designated node, is particularly practical and effective for community division of a larger network structure, has high efficiency and is equivalent to the global community discovery in effect, the relevant groups are rapidly discovered through the input nodes, the group screening is carried out by adding weak time association in group judgment, the group misjudgment rate can be effectively reduced, the local community effect is improved, the relevance in communities is strong but the non-community members can be reasonably pruned, and the accuracy of the community discovery is enhanced.
Drawings
FIG. 1 is a flow chart of the present invention.
Detailed Description
The invention will be further described with reference to the drawings and examples.
Example 1: the embodiment aims to provide a community discovery algorithm based on local paths, most of the existing community discovery algorithms are global, part of the local discovery algorithms are based on central nodes, and the core nodes of the local discovery algorithm are not high in model performance and low in efficiency, so that the requirements of people cannot be met.
The present invention is described in detail below with reference to an example, specifically as shown in fig. 1, and the present embodiment provides a local path-based community discovery algorithm, and uses a bank to discover abnormal transactions as an example, and specifically includes the following steps:
step 1, data acquisition
Acquiring a designated node from a database, and acquiring an associated node which has direct correlation or primary indirect correlation with the designated node; wherein the designated node is node detailed transaction flow data obtained by screening from a big data platform.
Step 2: data preprocessing
The method comprises the steps of obtaining an original relation between a designated node and an associated node, and performing deduplication on the original relation based on weak deduplication logic of time postductility and data equality to obtain an effective relation between the designated node and the associated node; for example, if the data is complete, the transaction of B < -, A exists, if the data is incomplete, even if the transaction of A < - > B exists for a plurality of times, the transaction of B < -, A does not necessarily correspond to the transaction of B < -, the time does not have the correspondence, the format is uniformly converted into A < - > B, a conversion mark is left, and the data is de-duplicated through the time and the conversion mark; then designating the number of times of the current and the total value of all the current and the total value in the effective relation determination time period as an edge index; the determined time may be one month, one quarter, or one year.
Step 3, calculating local similarity coefficients
If all the times of the node in the determined time period exceed the preset times y and the total value exceeds the preset value x, determining that the relationship between the node and the associated node is a, otherwise, b is b, and a+b=1; in this example a is 1 and b is 0, then the adjacency matrix A of the node network is obtained from the values of a and b ij ；
Calculating local similarity coefficients according to the adjacency matrix:
in the formula (1), lambda is the maximum characteristic value, A ij Is an adjacency matrix, S ij Is a local similarity coefficient;
local similarity coefficient in this example:
a in formula (2) ij K is a coefficient, and the value of k is decreased along with the path length; in the actual use process, S in the calculation ij The information obtained when the order of the (E) is the shortest path is free of repeated information, and the accuracy is good. When S is ij The order of (a) is not equal to the average shortest path (e.g. there is a direct relationship between ij, then S is employed ij ＝k 1 A ij Can better represent the local similarity coefficient by adoptingRepeated path computation thereof may result in a larger local similarity coefficient). An excessively high order will form information redundancy, and generate six degrees of separation, so three S are used in this example ij To calculate the local similarity coefficient.
Step 4, comparing the local similarity coefficient with a threshold value
Comparing the local similarity coefficient obtained in the step 3 with a set threshold, if the local similarity coefficient is larger than the set threshold, adding the node network into a community, otherwise, not adding; performing community expansion through local coefficient judgment until all nodes are divided;
step 5, obtaining the rationality of the community by adopting local modularity inspection
Defining node set of community as V, adding all adjacent nodes of these nodes into the set to form new set V, i.e. V is adjacent node of V, and V is adjacent matrix of V is
Positioning the local modularity, similar to the global modularity, the size of the proportion of the elements in the node set V, which are all included in the node set V, can be used to measure the quality of a community:
delta (c) in equation 4 i ，c j ) If i, j are both in V, then 1, otherwise 0; m is m * The number of edges within the adjacency matrix is identified.
If the modularity is larger than a preset modularity threshold, the community rationality is indicated to meet the requirements, otherwise, the next step is carried out; setting a threshold value as a community expansion cycle ending condition, ending a community discovery algorithm if the threshold value is met, otherwise, carrying out chain time weak correlation calculation to carry out community attenuation.
Step 6, processing communities with unqualified rationality
Ordering according to the degree of nodes in communities, and performing time weak association judgment on the node with the smallest degree, wherein the basis of the time weak association judgment is as follows: chain time weak correlation: if the chain is A- > B- > C, then A- > B occurs before B- > C.
If the weak association judgment meets the requirement, the node is kept in the community, the node is removed from the sequence, and the weak association judgment is carried out on the node with the minimum degree again; if the weak association judgment is not satisfactory, the node is removed from the community, and the modularity is recalculated and compared according to the mode in the step 5 until the modularity is greater than the modularity threshold.
In the step, the nodes are removed by using the chain time weak association, if the nodes are not in the chain time weak association, edge deletion operation is carried out on the data, and the local correlation relation between the nodes related to the deletion relation is recalculated until the local modularity is met, so that the accuracy of the data in communities is ensured, and the false judgment rate of the communities is effectively reduced.
The method and the system can analyze large-batch transaction data of banks, particularly analyze streaming funds, find abnormal transaction users, calculate local modularity of communities through a weak correlation chain, exclude correlation users which trade normally with a group organization, judge whether the correlation users are properly added into communities from peripheral accounts from correlation accounts, and finally prune part of accounts through the community modularity to achieve better community effect.
Therefore, the method adopts the gradual spreading type increase from the local designated nodes, particularly the community division on a larger network structure is very practical and effective, the discovery process is efficient and equivalent to the overall community discovery in effect, the relevant groups are rapidly discovered through the input nodes, the group screening is carried out by adding time weak association in the group judgment, the group misjudgment rate can be effectively reduced, the local community effect is improved, the relevance in communities is strong but the non-community members can be reasonably pruned, and the accuracy of the community discovery is enhanced.
Example 2: this embodiment is substantially the same as embodiment 1, except that: aiming at the defect of the existing similarity coefficient calculation, the triangular similarity coefficient replacement and supplement are added in the embodiment.
The specific scheme is as follows
Firstly, obtaining similarity coefficient based on adjacent matrix according to step 3
The core idea of obtaining the similarity coefficient based on the adjacency matrix is that the weighted sum of the path numbers of different lengths between i and j is adopted, but for non-high-density convergent communities, the path numbers of different lengths are single (chained communities), and S is obtained ij The value is small, and the integrity of communities is damaged.
The similarity coefficients are thus supplemented as follows:
Similar ij ＝max(S ij ,C j )
wherein C is j And calculating the cluster coefficient of the j node for the j node based on the triangle cluster coefficient.
Wherein v is i Is the vertex i, v j Is vertex j, e ij K, which are edges of vertices i and j j The number of nodes directly connected to the node j is indicated.
Using Similar ij The effect is far more remarkable than S in the aspect of facing the chain transaction structure community ij In the non-chain trade structure community, due to C j Considering only neighboring nodes and not considering relationships between indirect nodes so S in a non-chained transaction structure ij >C j 。
Will step by stepThe similarity coefficient calculation in step 3 is replaced by Similar ij The communities are not excessively pruned when facing the communities with the chain transaction structure.
Example 3: this embodiment is substantially the same as embodiment 1, except that: aiming at the defect of the existing similarity coefficient calculation, the embodiment adds the triangle similarity coefficient to make the missing supplement.
The specific scheme is as follows
Firstly, obtaining similarity coefficient based on adjacent matrix according to step 3
The core idea of the similarity coefficient based on the adjacency matrix is that the weighted sum of path numbers with different lengths between i and j is only the first three terms, and when three internal relations exist between i and j (the three relations are i- > x- > j, wherein x is any node), the path number weight with the distance from i to j exceeding 3 is not calculated, and the information is lost to a certain extent.
The similarity coefficients are thus supplemented as follows:
Similar ij ＝a*S ij +b*Jaccard ij
wherein Jaccard ij Is named as a similarity index of Jaccard, and the definition mode is that the common neighbor number of two vertexes is compared with the sum of all neighbor numbers of the vertexes;
wherein v is i Is a set of nodes directly related to vertex i, where v j Using Similar for a set of nodes directly related to vertex j ij The problem of information loss based on adjacency matrix is reduced to some extent.
Wherein a and b are weights, in this embodiment, the values a and b are both 0.5, i.e., similar ij ＝0.5*S ij +0.5*Jaccard ij 。
Claims (4)
1. A local path-based community discovery algorithm, characterized by: the method comprises the following steps:
step 1, data acquisition
Acquiring a designated node from a database, and acquiring an associated node which has direct correlation or primary indirect correlation with the designated node;
step 2, data preprocessing
The method comprises the steps of obtaining an original relation between a designated node and an associated node, and performing deduplication on the original relation based on weak deduplication logic of time postductility and data equality to obtain an effective relation between the designated node and the associated node; then designating the number of times of the node and the total value of all the times in the effective relation determination time period as an edge index;
step 3, calculating local similarity coefficients
If all the times of the node in the determined time period exceed the preset times y and the total value exceeds the preset value x, determining that the relationship between the node and the associated node is a, otherwise, b is b, and a+b=1; then obtaining the adjacency matrix of the node network in the determined time period according to the values of a and b；
Calculating local similarity coefficients according to the third-order adjacency matrix:
；
wherein lambda is the maximum eigenvalue,is an adjacency matrix->Is a local similarity coefficient, said +.>The representation being directly connected to node iNode number of->Representing the number of nodes directly connected with the node j;
step 4, comparing the local similarity coefficient with a threshold value
Comparing the local similarity coefficient obtained in the step 3 with a set threshold, if the local similarity coefficient is larger than the set threshold, adding the node network into a community, otherwise, not adding; performing community expansion through local coefficient judgment until all nodes are divided;
step 5, obtaining the rationality of the community by adopting local modularity inspection
The set of nodes defining the community is V,is the adjacent node of V,>is the adjacent matrix of (a)
；
Locating local modularity
；
Wherein the method comprises the steps ofIf i, j are both in V, then 1, otherwise 0;
identifying a number of edges within the adjacency matrix;
if the modularity is larger than a preset modularity threshold, the community rationality is indicated to meet the requirements, otherwise, the next step is carried out;
step 6, processing communities with unqualified rationality
Sorting according to the degree of nodes in communities, performing time weak association judgment on the node with the smallest degree, if the weak association judgment meets the requirement, keeping the node in communities, removing the node from the sorting, and performing weak association judgment on the node with the smallest degree again; if the weak association judgment is not satisfactory, the node is removed from the community, and the modularity is recalculated and compared according to the mode in the step 5 until the modularity is greater than the modularity threshold.
2. The local path-based community discovery algorithm of claim 1, wherein: in the step 1, the designated node is obtained through big data screening.
3. The local path-based community discovery algorithm of claim 1, wherein: in step 2, the determined time period is one month or one year.
4. The local path-based community discovery algorithm of claim 1, wherein: in step 3, a is 1 and b is 0.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011623050.7A CN112989272B (en) | 2020-12-31 | 2020-12-31 | Community discovery algorithm based on local path |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011623050.7A CN112989272B (en) | 2020-12-31 | 2020-12-31 | Community discovery algorithm based on local path |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112989272A CN112989272A (en) | 2021-06-18 |
| CN112989272B true CN112989272B (en) | 2024-02-27 |
Family
ID=76345158
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011623050.7A Active CN112989272B (en) | 2020-12-31 | 2020-12-31 | Community discovery algorithm based on local path |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112989272B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113420994B (en) * | 2021-06-28 | 2022-10-18 | 山东大学 | Method and system for evaluating structure flexibility of active power distribution network |
| CN113837879B (en) * | 2021-09-14 | 2023-12-19 | 上证所信息网络有限公司 | Abnormality detection method for index quotation |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107222334A (en) * | 2017-05-24 | 2017-09-29 | 南京大学 | Suitable for the local Combo discovering method based on core triangle of social networks |
| WO2019100967A1 (en) * | 2017-11-23 | 2019-05-31 | 中国银联股份有限公司 | Method and device for identifying social group having abnormal transaction activity |
| CN110084423A (en) * | 2019-04-24 | 2019-08-02 | 大连民族大学 | A kind of link prediction method based on local similarity |
| CN110781940A (en) * | 2019-10-17 | 2020-02-11 | 成都理工大学 | Fuzzy mathematics-based community discovery information processing method and system |
| CN111008215A (en) * | 2019-11-29 | 2020-04-14 | 中科院计算技术研究所大数据研究院 | Expert recommendation method combining label construction and community relation avoidance |
| CN111030854A (en) * | 2019-12-04 | 2020-04-17 | 兰州交通大学 | Complex network community discovery method under Spark cloud service environment |
| WO2020253222A1 (en) * | 2019-06-19 | 2020-12-24 | 江南大学 | Community detection method for dynamic residue interaction network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10846052B2 (en) * | 2016-10-27 | 2020-11-24 | Tencent Technology (Shenzhen) Company Limited | Community discovery method, device, server and computer storage medium |
-
2020
- 2020-12-31 CN CN202011623050.7A patent/CN112989272B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107222334A (en) * | 2017-05-24 | 2017-09-29 | 南京大学 | Suitable for the local Combo discovering method based on core triangle of social networks |
| WO2019100967A1 (en) * | 2017-11-23 | 2019-05-31 | 中国银联股份有限公司 | Method and device for identifying social group having abnormal transaction activity |
| CN110084423A (en) * | 2019-04-24 | 2019-08-02 | 大连民族大学 | A kind of link prediction method based on local similarity |
| WO2020253222A1 (en) * | 2019-06-19 | 2020-12-24 | 江南大学 | Community detection method for dynamic residue interaction network |
| CN110781940A (en) * | 2019-10-17 | 2020-02-11 | 成都理工大学 | Fuzzy mathematics-based community discovery information processing method and system |
| CN111008215A (en) * | 2019-11-29 | 2020-04-14 | 中科院计算技术研究所大数据研究院 | Expert recommendation method combining label construction and community relation avoidance |
| CN111030854A (en) * | 2019-12-04 | 2020-04-17 | 兰州交通大学 | Complex network community discovery method under Spark cloud service environment |
Non-Patent Citations (5)
| Title |
|---|
| 基于局部路径的社团发现算法;高红艳;刘飞;;电气自动化(第06期);全文 * |
| 基于群体智能的自组织重叠社团结构分析算法;孙韩林;马素刚;王忠民;;计算机应用研究(第05期);全文 * |
| 基于节点内聚系数的局部社团发现算法;赵文涛;赵好好;孟令军;;计算机应用与软件(第12期);全文 * |
| 基于节点相似度的社会网络社团发现的算法研究;李超男;硕士电子期刊(第1期);全文 * |
| 高红艳，刘飞.基于局部路径的社团发现算法.电气自动化.2014,第36卷(第6期),全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112989272A (en) | 2021-06-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107482626B (en) | Method for identifying key nodes of regional power grid | |
| WO2019238109A1 (en) | Fault root cause analysis method and apparatus | |
| WO2021073462A1 (en) | 10 kv static load model parameter identification method based on similar daily load curves | |
| CN112989272B (en) | Community discovery algorithm based on local path | |
| CN103476051B (en) | A kind of communication net node importance evaluation method | |
| CN110417011B (en) | Online dynamic security assessment method based on mutual information and iterative random forest | |
| CN112149873A (en) | Low-voltage transformer area line loss reasonable interval prediction method based on deep learning | |
| CN115618249A (en) | Low-voltage power distribution station area phase identification method based on LargeVis dimension reduction and DBSCAN clustering | |
| CN111178957B (en) | Method for early warning sudden increase of electric quantity of electricity consumption customer | |
| CN111709668A (en) | Power grid equipment parameter risk identification method and device based on data mining technology | |
| Zhang et al. | A hypothesis testing framework for modularity based network community detection | |
| CN109858822B (en) | Information power fusion system reliability assessment method based on flow correlation analysis | |
| CN114971345B (en) | Quality measuring method, equipment and storage medium for built environment | |
| CN112989526B (en) | Aviation network key node identification method based on kernel extreme learning machine | |
| CN116151799A (en) | BP neural network-based distribution line multi-working-condition fault rate rapid assessment method | |
| CN115409317A (en) | Transformer area line loss detection method and device based on feature selection and machine learning | |
| CN111931861B (en) | Anomaly detection method for heterogeneous data set and computer-readable storage medium | |
| CN114329867A (en) | Scale-free network robustness measuring method based on motif | |
| CN112926664A (en) | Feature selection and CART forest short-time strong rainfall forecasting method based on evolutionary algorithm | |
| CN111092425B (en) | Net rack clustering analysis method and system based on topological characteristics | |
| CN111652102A (en) | Power transmission channel target object identification method and system | |
| CN114928545B (en) | Spark-based large-scale flow data key node calculation method | |
| CN117522181A (en) | Regional toughness assessment method and device based on functional connection | |
| CN117852244A (en) | Power grid model light-weight method and system based on node aggregation | |
| Wang et al. | Visualization of the uk stock market based on complex networks for company’s revenue forecast |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |