CN102609528A - Frequent mode association sorting method based on probabilistic graphical model - Google Patents
Frequent mode association sorting method based on probabilistic graphical model Download PDFInfo
- Publication number
- CN102609528A CN102609528A CN2012100316626A CN201210031662A CN102609528A CN 102609528 A CN102609528 A CN 102609528A CN 2012100316626 A CN2012100316626 A CN 2012100316626A CN 201210031662 A CN201210031662 A CN 201210031662A CN 102609528 A CN102609528 A CN 102609528A
- Authority
- CN
- China
- Prior art keywords
- frequent
- frequent item
- big
- mode
- keyword
- 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.)
- Granted
Links
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a frequent mode association sorting method based on a probabilistic graphical model, and provides a frequent mode mutual relation representation and frequent mode association sorting method based on the probabilistic graphical model based on execution results of Apriori frequent pattern mining algorithms. A Markov network which is an important probabilistic graphical model is used as a basic framework for knowledge representation to set up internal relation between the frequent mode and the probabilistic graphical model and build the Markov network included in the frequent mode, the frequent mode is subjected to association sorting of different abstract hierarchies by node aggregations from bottom to top, mutual relations of any forms among the frequent modes can be conveniently and efficiently expressed in the global view, better flexibility of association sorting of the users in different abstract hierarchies is achieved, and theoretical basis and technical foundation are provided for subsequent development.
Description
One, technical field: the invention discloses a kind of frequent mode associative classification method, relate to a kind of based on the expression of mutual relationship between the frequent mode of probability graph model and on different abstraction hierarchies, carry out the method for associative classification based on probability graph model (Probabilistic Graphical Model).Belong to data mining and technical field of information processing.
Two, background technology
Data object in the reality, except attribute itself, the behavior of object and the mutual relationship that produces owing to behavior also are the important evidence that it is classified.Utilize Mining Algorithms of Frequent Patterns frequently to be appeared at the pattern of data centralization; Utilize correlation rule to express the mutual relationship between the frequent mode; Classical sorting algorithm with the attribute of object itself be the basis, the mutual relationship that produces owing to behavior between the object of consideration; For this reason, the correlation rule of mutual relationship between indicated object is used for classification of Data, known associative classification method is carried out the classification analysis of data object based on correlation rule.Dong Jie (Dalian University of Technology's PhD dissertation, 2009) has proposed a kind of association rule mining and associative classification algorithm based on bit table; Chen Guoqing etc. (< management of information resources journal >, 2011 (2)) have introduced the associative classification method based on information entropy; (< computer research and development >, 2011,48 (4): 567-575) proposed a kind of Fuzzy Correlation sorting technique such as Huo Weigang based on the multi-target evolution algorithm.
Basic technology means as associative classification; The correlation rule method for expressing of frequent mode can not be from mutual relationship complicated between the angle effective expression frequent mode of the overall situation; The overall probability distribution of related frequent mode and the uncertainty of mutual relationship can not be described; For this reason, known method is utilized graph model expansion frequent mode and Mining Association Rules algorithm.(< computer integrated manufacturing system >, 2008,14 (6): 1220-1229) proposed a kind of Mining Algorithms of Frequent Patterns such as Geng Runian based on overall situation figure traversal; (< computer engineering >, 2010,36 (13): 9-6) proposed a kind of weighted association rules model, and utilized associated diagram storage frequent mode collection such as Chen Wen based on associated diagram; Hu Chunling etc. (< software journal >; 2011; 22 (12): 2934-2950) proposed a kind of frequent mode interest-degree and calculated and the beta pruning strategy, and effectively utilized the reasoning algorithm of Bayesian network to come the regular support of compute associations based on this probability graph model of Bayesian network.
Known relatively frequent mode method for expressing; Can represent the overall mutual relationship of arbitrary form between the frequent mode and the uncertainty of mutual relationship based on probability graph model; Analyze the tightness degree of mutual relationship between frequent mode and carry out the merging of node based on probability graph model, can carry out the frequent mode classification at different abstraction hierarchies.With the cause-effect relationship between the frequent mode is starting point; The probability graph model method for expressing of frequent mode has been proposed; Having set up the equivalence from the frequent mode to the probability graph model changes the mechanism; Provided frequent mode level method for congregating, used it in the problem of automatic classification of scientific paper and paper author contact, had higher efficient and classification accuracy based on probability graph model character.The method can realize that easily and efficiently the overall situation of relation of interdependence is represented between the frequent mode with a unified model; Can satisfy different abstraction hierarchy users' associative classification demand; Has retractility preferably, for follow-up research and development provide theoretical foundation and technical foundation.
Three, summary of the invention
The present invention provides a kind of frequent mode associative classification method based on probability graph model.On the execution result of Apriori Mining Algorithms of Frequent Patterns, provide a kind of based on the expression of mutual relationship between the frequent mode of probability graph model and the associative classification method of frequent mode.With the basic framework of this important probability graph model of markov net (Markov network) as the representation of knowledge; Set up the inner link of frequent mode and probability graph model; Make up the markov net that contains in the frequent mode, frequent mode is carried out the associative classification on the different abstraction hierarchies through the bottom-up gathering of node.Can represent the mutual relationship of arbitrary form between frequent mode from the angle of the overall situation easily and efficiently, different abstraction hierarchy users' associative classification has retractility preferably, for follow-up research and development provide theoretical foundation and technical foundation.
The present invention gathers completion by the following step
Technological process of the present invention is: at first, based on the Apriori Mining Algorithms of Frequent Patterns, support is set, obtain very big Frequent Item Sets; Then; Each very big Frequent Item Sets is made up initial non-directed graph respectively, and carry out the merging of initial non-directed graph according to the public project collection between them, and then the condition independence between the node in the test pattern; The deletion condition is the limit independently, obtains frequent item and concentrates the markov net that contains; Then, the markov net that obtains is carried out the string processing, the markov net of stringization is expressed as threaded tree,, thereby obtain the preliminary classification of frequent mode with the summit of a string beggar figure as threaded tree; Further, merging is assembled on the summit of threaded tree, obtained reflecting the classification of higher abstraction hierarchy, till meeting consumers' demand with bottom-up mode.
(1) obtain frequent mode: based on the Apriori Mining Algorithms of Frequent Patterns, and the support threshold value is set, obtains the 1-frequent item set, the 2-frequent item set ..., till can not obtaining bigger frequent item set, thereby obtain very big frequent item set.
Based on the Apriori Mining Algorithms of Frequent Patterns, to item collection I={i
1..., i
n, support threshold epsilon (0＜ε＜1) is set, if the subset X of I satisfies probability P (X)>=ε, then X is a frequent item set.At first obtain containing the 1-Frequent Item Sets of 1 item, obtain containing the 2-Frequent Item Sets of 2 items again ..., carry out successively, till can not obtaining bigger frequent item set.Thereby obtain very big Frequent Item Sets;
(2) make up the markov net that contains in the frequent mode: to each very big Frequent Item Sets; At first make up with each frequent item wherein as the full-mesh non-directed graph of node; Again the pairing complete subgraph of each very big Frequent Item Sets is merged; Then according to whether condition independently comes to confirm the deletion and the reservation on limit between the frequent item, thereby obtain reacting between the frequent item overall situation markov net that is mutually related.
1. each very big Frequent Item Sets is made up non-directed graph respectively: to very big frequent item set A
i, the node as figure with wherein connects A with nonoriented edge
iIn any two different items, obtain A
iCorresponding full-mesh non-directed graph G (A
i), like Fig. 2, Fig. 3 and shown in Figure 4;
2. merge the corresponding non-directed graph of all frequent item sets: for any two A that have public keys
iAnd A
j, with A
iIn each and A
jIn other link to each other with nonoriented edge, thereby the non-directed graph that each very big frequent item set is corresponding merges, and obtains overall non-directed graph G, as shown in Figure 5;
3. leave out the corresponding limit of condition independent processing unit, obtain the markov net:
Use<α | Z| β>Expression " α and β condition are independent of Z ", if P (α, Z, β)=P
λ(α, Z) P
λ(β, Z)/P
λ(Z), wherein
X is a frequent item set, and λ is given probability threshold value.
If X is very big frequent item set, α, β ∈ X, have α | the X-alpha-beta | β>the total establishment.For the corresponding non-directed graph of all frequent item sets, examination G (A
i) in any nonoriented edge (a
Il, a
Ik), if<a
Il| A
i-a
Il-a
Ik| a
Ik>Setting up (is a
IlWith a
IkCondition is independent of G (A
i) in other nodes), then from G, delete limit (a
Il, a
Ik); If a
IlAnd a
IkBe again A
jIn frequent and<a
Il| A
i-a
Il-a
Ik| a
Ik>Setting up (is a
IlWith a
IkCondition is independent of G (A
j) in other nodes), then also from G, delete limit (a
Il, a
Ik).Thereby set up the relation between frequent mode and the condition independence; Obtained the non-directed graph structure of relation of interdependence between the frequent item of expression; This graph structure satisfy probability graph model necessary condition, be effective frequent markov net; Be referred to as a related markov net (Item Association Markov Network), as shown in Figure 6.
(3) level of frequent mode is assembled: according to the definition of stringization; (non-directed graph is called string figure; Arbitrary length all has a string at least greater than 3 ring in figure), with the markov net string processing that makes up, set up the acyclic preface of the very big complete subgraph of each node in the markov net simultaneously; And then to obtain with very big complete subgraph be the connection tree of node; The gathering that connects node in the tree according to the acyclic preface that connects very big complete subgraph in the tree merges, and bottom-up mode repeats this process, till satisfying the required level of abstraction of user.
1. get in touch criterion closely with stringization (Chordal) as frequent item; Obtain the preface of the related markov net of item and the string beggar figure of stringization: based on the notion of non-directed graph stringization; Each length is no less than 4 ring, and all to carry out stringization (be trigonometric ratio; Make the length of each ring be not more than 3), each length is no more than node in 3 the ring and constitutes a string beggar and scheme x
i, each string beggar figure comprises frequent closely of contact and corresponding initial class, and is as shown in Figure 7; Further obtain the preface (x of string beggar figure based on following standard
1..., x
m), for the class that obtains higher abstraction hierarchy lays the foundation:
1≤j≤i wherein;
2. the string non-directed graph is expressed as threaded tree (Join Tree): the markov net of stringization can be described with tree construction, is called threaded tree; And threaded tree itself is a stringization, has comprised frequent closely of contact.The string beggar is schemed as the summit, if C
iWith C
jThere is public frequent, then C
iWith C
jBetween a nonoriented edge is arranged, obtain threaded tree T, as shown in Figure 8;
3. the threaded tree node is assembled merging; Realize the frequent mode associative classification of different abstraction hierarchies: according to the preface of string beggar figure; Summit and head end node that each nonoriented edge among the threaded tree T is terminal merge; Obtain new threaded tree T ', a class of the higher abstraction hierarchy of each node correspondence wherein is like Fig. 9 and shown in Figure 10.Repeat this process with bottom-up mode, obtain increasing class, till satisfying the required level of abstraction of user.
Compare advantage and good effect that the present invention has with known technology
(1) through making up probability graph model; With a unified model, described the mutual relationship between the frequent mode from the angle of the overall situation; It is the expansion of frequent mode and association rule mining method; More easily realize the modeling of arbitrary form mutual relationship between frequent mode, remedied the deficiency of representing mechanism based on mutual relationship between the frequent mode of correlation rule.
(2) be starting point with the cause-effect relationship between frequent mode, set up that the equivalence from the frequent mode to the probability graph model is changed the mechanism, the expression of frequent mode joint probability distribution mechanism, reflected complementary uncertainty between frequent mode quantitatively.
(3) assemble based on the node of probability graph model and realize associative classification; Owing to only consider the classification that local correlations is brought or the one-sidedness and the inaccuracy of cluster result, the easy implementation of associative classification and result's correctness have been improved when having avoided carrying out associative classification based on correlation rule; Realize the associative classification of the different abstraction hierarchies of frequent mode, had better scalability, can satisfy user's different demands.
(4) ripe probability graph model inference method can be the support technology that associative classification provides quantitative analysis and calculating, for solving the auto-associating classification and the hot issue that needs to be resolved hurrily at present such as calculating based on the society of associative classification strong technical support is provided.
Four, description of drawings
Fig. 1 technology path figure of the present invention.Comprise following three major parts: obtain frequent mode (pre-service), make up probability graph model and hierarchical associated classification;
Fig. 2, Fig. 3 and Fig. 4 are respectively the corresponding initial non-directed graph of three Frequent Item Sets:
The undirected subgraph of Fig. 2 full-mesh 1..Node is very big frequent item set (A, B, in C) frequent;
The undirected subgraph of Fig. 3 full-mesh 2..Node is very big frequent item set (C, in D) frequent;
The undirected subgraph of Fig. 4 full-mesh 3..Node is very big frequent item set (D, E, in F) frequent;
All non-directed graphs of frequent of Fig. 5.Combined diagram 2, Fig. 3 and Fig. 4 obtain Fig. 5, node be all frequent item set U=(A, B, C, D, E, in F) frequent, the limit of adding when merging the undirected subgraph of full-mesh is represented with two-wire;
The related markov net of item of Fig. 6 keyword Frequent Item Sets U.Fig. 5 is carried out obtaining after the condition independent test;
The related markov net G of item of Fig. 7 stringization.Fig. 6 is carried out the string processing obtain, wherein x
1=" frequent ", x
2=" Apriori ", x
3=" beta pruning ", x
4=" classification ", x
5=" Bayesian network ", x
6=" group tree ";
The threaded tree T of the related markov net G of item of Fig. 8 stringization
1C wherein
1=(x
1, x
2, x
3) representative " correlation rule ", C
2=(x
2, x
3, x
5) representative " graph model excavation ", C
3=(x
2, x
4) representative " classification analysis ", C
4=(x
5, x
6) representative " probability graph model ";
The connection tree graph T that Fig. 9 is new
2To T among Fig. 8
1The summit assemble to merge and to obtain, wherein
Representative " association rule mining ",
Representative " uncertain Knowledge Discovery ",
Representative " associative classification ";
The threaded tree T that Figure 10 is new
3The threaded tree T of high abstraction hierarchy
4Respectively to T
2And T
3The summit assemble to merge and to obtain, wherein
Representative " artificial intelligence ",
Representative " data mining ";
Expression " data and knowledge engineering ".
Five, embodiment
Embodiment 1: scientific paper keyword associative classification
(1) Item Sets: extracting keywords from the scientific paper of delivering (Keywords) is also added up respectively the frequent degree that each speech occurs, if two keywords appear in the same piece of writing paper, representes then that support that both keyword occurs is simultaneously calculated and adds 1;
(2) very big Frequent Item Sets: the minimum support threshold value is set, uses the Apriori algorithm, scanning keyword and counting; Obtain the set of 1-Frequent Item Sets; Further obtain the set of 2-Frequent Item Sets ..., constantly carry out till can not finding the k-Frequent Item Sets again;
(3) to the very big Frequent Item Sets of each keyword, at first make up so that wherein each frequent item is as the full-mesh non-directed graph of node, whether condition independently comes to confirm the deletion and the reservation on limit between frequent then, thereby obtains the subgraph of each very big Frequent Item Sets; Again the corresponding subgraph of each very big Frequent Item Sets is merged, obtain reflecting the markov net of overall mutual relationship between the frequent item, U=(A, B; C, D, E F) is the 1-Frequent Item Sets of keyword; At first obtain respectively like Fig. 2, Fig. 3 and 3 undirected subgraphs of full-mesh shown in Figure 4, the common node according to each subgraph merges this 3 sub-graphs again, obtains corresponding to all non-directed graphs of frequent among the U, and is as shown in Figure 5; The keyword frequent item is carried out the condition independent test,, then leave out corresponding limit, (A if condition is independent; E), (A, F), (B, E) with (B, F) limit between these 4 pairs of nodes does not exist; For the non-directed graph among Fig. 5, < E|C, D|F>(being that E and F condition are independent of C and D); Then leave out the limit between E and the F, obtain the related markov net of item of keyword Frequent Item Sets U, as shown in Figure 6;
(4) if the related markov net of item of stringization is as shown in Figure 7, according to the preface (C of string beggar figure
1, C
2, C
3, C
4), obtain threaded tree T
1, as shown in Figure 8, the acyclic preface of the very big complete subgraph in threaded tree summit is (C among Fig. 8
1', C
2', C
3'), then to T
1In the summit assemble merging, obtain new, as to describe higher abstraction hierarchy keyword frequent item classification threaded tree T
2, as shown in Figure 9.To T
2In the summit assemble merging, obtain new threaded tree T
3, and then obtain C
1(the C of " '=
1", C
2") promptly obtains the threaded tree T of the highest abstraction hierarchy class
4, shown in figure 10.
Performance: 400 pieces of the scientific papers in the selection ScienceDirect database in 5 " theme (Subject) "; Choose 1500 keywords wherein; Write down these papers theme and under sub-topics information; Carry out above step (1)～(4), make up a related markov net from 1000 frequent and only need 15 milliseconds, obtain said sub-topics of paper and upper level subject classification information; Abstraction hierarchy in these two classification is compared for said type with paper itself respectively, and the error that this research institute gets the result is respectively 2.5% and 1.2%.
Claims (2)
1. frequent mode associative classification method based on probability graph model, it is characterized in that: it is accomplished according to the following steps,
(1) obtain frequent mode: based on the Apriori Mining Algorithms of Frequent Patterns, and the support threshold value is set, obtains the 1-frequent item set, the 2-frequent item set ..., till can not obtaining bigger frequent item set, thereby obtain very big frequent item set;
(2) make up the markov net that contains in the frequent mode: to each very big Frequent Item Sets; At first make up with each frequent item wherein as the full-mesh non-directed graph of node; Again the pairing complete subgraph of each very big Frequent Item Sets is merged; Then based on whether condition independently comes to confirm the deletion and the reservation on limit between the frequent item, thereby obtain reacting between the frequent item overall situation markov net that is mutually related;
(3) level of frequent mode is assembled: according to the definition of stringization; With the markov net string processing that makes up; Set up the acyclic preface of the very big complete subgraph of each node in the markov net simultaneously, and then to obtain with very big complete subgraph be the connection tree of node, merge according to connecting in the tree gathering that the very big acyclic preface of complete subgraph connects node in the tree; Bottom-up mode repeats this process, till satisfying the required level of abstraction of user.
2. the frequent mode associative classification method based on probability graph model according to claim 1 is characterized in that: a kind of scientific paper keyword associative classification method is accomplished according to the following steps,
(1) Item Sets: extracting keywords from the scientific paper of delivering (Keywords) is also added up respectively the frequent degree that each speech occurs, if two keywords appear in the same piece of writing paper, representes then that support that both keyword occurs is simultaneously calculated and adds 1;
(2) very big Frequent Item Sets: the minimum support threshold value is set, uses the Apriori algorithm, scanning keyword and counting; Obtain the set of 1-Frequent Item Sets; Further obtain the set of 2-Frequent Item Sets ..., constantly carry out till can not finding the k-Frequent Item Sets again;
(3) to the very big Frequent Item Sets of each keyword, at first make up so that wherein each frequent item is as the full-mesh non-directed graph of node, whether condition independently comes to confirm the deletion and the reservation on limit between frequent then, thereby obtains the subgraph of each very big Frequent Item Sets; Again the corresponding subgraph of each very big Frequent Item Sets is merged, obtain reflecting the markov net of overall mutual relationship between the frequent item, U=A, B; C, D, E, F are the 1-Frequent Item Sets of keyword; At first obtain 3 undirected subgraphs of full-mesh, the common node according to each subgraph merges this 3 sub-graphs again, obtains corresponding to all non-directed graphs of frequent among the U keyword frequent item being carried out the condition independent test; If condition is independent, then leave out corresponding limit, (A, E), (A; F), (B, E) with (B, F) limit between these 4 pairs of nodes does not exist, for the non-directed graph among Fig. 5; < E|C, D|F>then leaves out the limit between E and the F, obtains the related markov net of item of keyword Frequent Item Sets U;
(4) the preface C that schemes according to the string beggar
1, C
2, C
3, C
4, obtain threaded tree T
1, the acyclic preface of the very big complete subgraph in threaded tree summit does among Fig. 8
Then to T
1In the summit assemble merging, obtain new, as to describe higher abstraction hierarchy keyword frequent item classification threaded tree T
2, to T
2In the summit assemble merging, obtain new threaded tree T
3, and then obtain C
1The C of " '=
1", C
2", promptly obtain the threaded tree T of the highest abstraction hierarchy class
4
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210031662.6A CN102609528B (en) | 2012-02-14 | 2012-02-14 | Frequent mode association sorting method based on probabilistic graphical model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210031662.6A CN102609528B (en) | 2012-02-14 | 2012-02-14 | Frequent mode association sorting method based on probabilistic graphical model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102609528A true CN102609528A (en) | 2012-07-25 |
| CN102609528B CN102609528B (en) | 2014-06-18 |
Family
ID=46526900
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210031662.6A Expired - Fee Related CN102609528B (en) | 2012-02-14 | 2012-02-14 | Frequent mode association sorting method based on probabilistic graphical model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102609528B (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015077942A1 (en) * | 2013-11-27 | 2015-06-04 | Hewlett-Packard Development Company, L.P. | Relationship extraction |
| CN106033449A (en) * | 2015-03-17 | 2016-10-19 | 哈尔滨工业大学深圳研究生院 | Method and device for item-sets mining |
| CN106649479A (en) * | 2016-09-29 | 2017-05-10 | 国网山东省电力公司电力科学研究院 | Probability graph-based transformer state association rule mining method |
| CN107247995A (en) * | 2016-09-29 | 2017-10-13 | 上海交通大学 | Transmission line of electricity running status association rule mining and Forecasting Methodology based on Bayesian model |
| CN108132927A (en) * | 2017-12-07 | 2018-06-08 | 西北师范大学 | A kind of fusion graph structure and the associated keyword extracting method of node |
| CN108228607A (en) * | 2016-12-14 | 2018-06-29 | 中国航空工业集团公司西安航空计算技术研究所 | Maximum frequent itemsets method for digging based on degree of communication |
| WO2019028710A1 (en) * | 2017-08-09 | 2019-02-14 | 深圳清华大学研究院 | Method for calculating support of candidate item set on basis of graphic structure data, and application thereof |
| CN111400432A (en) * | 2020-06-04 | 2020-07-10 | 腾讯科技（深圳）有限公司 | Event type information processing method, event type identification method and device |
| CN112148960A (en) * | 2019-06-27 | 2020-12-29 | 北京百度网讯科技有限公司 | Method, device, equipment and storage medium for determining category of attention point |
| CN113094746A (en) * | 2021-03-31 | 2021-07-09 | 北京邮电大学 | High-dimensional data publishing method based on localized differential privacy and related equipment |
| CN114142931A (en) * | 2021-12-13 | 2022-03-04 | 北京邮电大学 | Complex channel communication method based on BIC-DAF-MOEA |
| CN116721001A (en) * | 2023-08-10 | 2023-09-08 | 江苏网进科技股份有限公司 | Smart city resource management method based on digital twinning |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070239694A1 (en) * | 2006-02-27 | 2007-10-11 | Singh Ambuj K | Graph querying, graph motif mining and the discovery of clusters |
| CN101446978A (en) * | 2008-12-11 | 2009-06-03 | 南京大学 | Core node discovery method based on frequent itemset mining |
-
2012
- 2012-02-14 CN CN201210031662.6A patent/CN102609528B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070239694A1 (en) * | 2006-02-27 | 2007-10-11 | Singh Ambuj K | Graph querying, graph motif mining and the discovery of clusters |
| CN101446978A (en) * | 2008-12-11 | 2009-06-03 | 南京大学 | Core node discovery method based on frequent itemset mining |
Non-Patent Citations (1)
| Title |
|---|
| 王元珍等: "基于关联规则挖掘的中文文本自动分类", 《小型微型计算机***》 * |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015077942A1 (en) * | 2013-11-27 | 2015-06-04 | Hewlett-Packard Development Company, L.P. | Relationship extraction |
| US10643145B2 (en) | 2013-11-27 | 2020-05-05 | Micro Focus Llc | Relationship extraction |
| CN106033449A (en) * | 2015-03-17 | 2016-10-19 | 哈尔滨工业大学深圳研究生院 | Method and device for item-sets mining |
| CN106649479A (en) * | 2016-09-29 | 2017-05-10 | 国网山东省电力公司电力科学研究院 | Probability graph-based transformer state association rule mining method |
| CN107247995A (en) * | 2016-09-29 | 2017-10-13 | 上海交通大学 | Transmission line of electricity running status association rule mining and Forecasting Methodology based on Bayesian model |
| CN106649479B (en) * | 2016-09-29 | 2020-05-12 | 国网山东省电力公司电力科学研究院 | Transformer state association rule mining method based on probability graph |
| CN108228607A (en) * | 2016-12-14 | 2018-06-29 | 中国航空工业集团公司西安航空计算技术研究所 | Maximum frequent itemsets method for digging based on degree of communication |
| CN108228607B (en) * | 2016-12-14 | 2021-10-15 | 中国航空工业集团公司西安航空计算技术研究所 | Maximum frequent item set mining method based on connectivity |
| US10776372B2 (en) | 2017-08-09 | 2020-09-15 | Research Institute Of Tsinghua University In Shenzhen | Method for computing support of itemset candidate based on graph structure data and application thereof |
| WO2019028710A1 (en) * | 2017-08-09 | 2019-02-14 | 深圳清华大学研究院 | Method for calculating support of candidate item set on basis of graphic structure data, and application thereof |
| CN108132927A (en) * | 2017-12-07 | 2018-06-08 | 西北师范大学 | A kind of fusion graph structure and the associated keyword extracting method of node |
| CN112148960A (en) * | 2019-06-27 | 2020-12-29 | 北京百度网讯科技有限公司 | Method, device, equipment and storage medium for determining category of attention point |
| CN112148960B (en) * | 2019-06-27 | 2024-03-22 | 北京百度网讯科技有限公司 | Method, device, equipment and storage medium for determining category of attention point |
| CN111400432A (en) * | 2020-06-04 | 2020-07-10 | 腾讯科技（深圳）有限公司 | Event type information processing method, event type identification method and device |
| CN113094746A (en) * | 2021-03-31 | 2021-07-09 | 北京邮电大学 | High-dimensional data publishing method based on localized differential privacy and related equipment |
| CN114142931A (en) * | 2021-12-13 | 2022-03-04 | 北京邮电大学 | Complex channel communication method based on BIC-DAF-MOEA |
| CN114142931B (en) * | 2021-12-13 | 2023-09-12 | 北京邮电大学 | Complex channel communication method based on BIC-DAF-MOEA |
| CN116721001A (en) * | 2023-08-10 | 2023-09-08 | 江苏网进科技股份有限公司 | Smart city resource management method based on digital twinning |
| CN116721001B (en) * | 2023-08-10 | 2023-11-17 | 江苏网进科技股份有限公司 | Smart city resource management method based on digital twinning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102609528B (en) | 2014-06-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102609528B (en) | Frequent mode association sorting method based on probabilistic graphical model | |
| Tagarelli et al. | Ensemble-based community detection in multilayer networks | |
| Moosavi et al. | Community detection in social networks using user frequent pattern mining | |
| Braun et al. | Effectively and efficiently mining frequent patterns from dense graph streams on disk | |
| JP6183376B2 (en) | Index generation apparatus and method, search apparatus, and search method | |
| Adcock et al. | Tree decompositions and social graphs | |
| Pan et al. | Clustering of designers based on building information modeling event logs | |
| CN107273934A (en) | A kind of figure clustering method merged based on attribute | |
| Qiao et al. | Unsupervised author disambiguation using heterogeneous graph convolutional network embedding | |
| Obaid et al. | Semantic web and web page clustering algorithms: a landscape view | |
| JPWO2016006276A1 (en) | Index generation apparatus and index generation method | |
| CN108960335A (en) | One kind carrying out efficient clustering method based on large scale network | |
| Shekhawat et al. | A classification technique using associative classification | |
| Lu et al. | A unified link prediction framework for predicting arbitrary relations in heterogeneous academic networks | |
| Wang et al. | Detecting overlapping communities in location-based social networks | |
| Tang et al. | A multi-resolution approach to learning with overlapping communities | |
| CN116450938A (en) | Work order recommendation realization method and system based on map | |
| Hao et al. | The research and analysis in decision tree algorithm based on C4. 5 algorithm | |
| Liu et al. | Graph-based soft-balanced fuzzy clustering | |
| Ning et al. | An adaptive node embedding framework for multiplex networks | |
| Zhou et al. | Identifying technology evolution pathways by integrating citation network and text mining | |
| Singh et al. | RSTDB a new candidate generation and test algorithm for frequent pattern mining | |
| Xu | Deep mining method for high-dimensional big data based on association rule | |
| Chan et al. | ciForager: Incrementally discovering regions of correlated change in evolving graphs | |
| Hristoskova et al. | A graph-based disambiguation approach for construction of an expert repository from public online sources |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20140618 Termination date: 20160214 |