CN110096900A - A kind of Frequent Pattern Mining method of efficient difference secret protection - Google Patents
A kind of Frequent Pattern Mining method of efficient difference secret protection Download PDFInfo
- Publication number
- CN110096900A CN110096900A CN201910363516.5A CN201910363516A CN110096900A CN 110096900 A CN110096900 A CN 110096900A CN 201910363516 A CN201910363516 A CN 201910363516A CN 110096900 A CN110096900 A CN 110096900A
- Authority
- CN
- China
- Prior art keywords
- frequent
- cset
- mode
- rank
- candidate
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000005065 mining Methods 0.000 title claims abstract description 30
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 47
- 230000007246 mechanism Effects 0.000 claims abstract description 10
- 238000005457 optimization Methods 0.000 claims description 10
- 230000006835 compression Effects 0.000 claims description 8
- 238000007906 compression Methods 0.000 claims description 8
- 238000012545 processing Methods 0.000 claims description 5
- 238000013480 data collection Methods 0.000 claims description 4
- 238000002474 experimental method Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 7
- 238000012360 testing method Methods 0.000 description 7
- 238000004458 analytical method Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 101150073719 Sh3kbp1 gene Proteins 0.000 description 2
- 238000009412 basement excavation Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2465—Query processing support for facilitating data mining operations in structured databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/60—Protecting data
- G06F21/62—Protecting access to data via a platform, e.g. using keys or access control rules
- G06F21/6218—Protecting access to data via a platform, e.g. using keys or access control rules to a system of files or objects, e.g. local or distributed file system or database
- G06F21/6245—Protecting personal data, e.g. for financial or medical purposes
Abstract
The invention discloses a kind of Frequent Pattern Mining methods of efficient difference secret protection, first input transaction data set (TDS) D, privacy budget ε, frequent mode number k, minimum support threshold value min_sup, and privacy budget ε is split as ε1、ε2、ε3, and meet ε=ε1+ε2+ε3;Set Cset, Sset, Pset are emptied respectively, wherein Cset is candidate frequent item set, and the set that the top-k frequent mode that Sset index mechanism is picked out is formed, Pset is the frequent mode set added after noise;Then truncation, l before only retaining are carried out to transaction data set (TDS) Dopt, on the other hand, after all frequent mode Candidate Sets are excavated in FP-Growth algorithm, frequent mode concept is closed in introducing, about subtracts Candidate Set scale.The present invention solves the problems, such as that information leakage is serious during Frequent Pattern Mining in big data application existing in the prior art.
Description
Technical field
The invention belongs to field of information security technology, and in particular to a kind of frequent mode digging of efficient difference secret protection
Pick method.
Background technique
With the high speed development of electronic information technology and the arriving of big data era, storage, collection, analysis and the hair of data
Cloth demand is increasing, by carrying out the letter that Information extraction and analysis can allow people to obtain more real worlds to these data
Breath.This kind of demand largely promotes the shared of data information, analysis and publication.Although just to the analyses of this kind of data
Benefit people's lives bring relevant benefit for the holder of these data, but a serious problem but highlights,
That is the privacy of user receives great threat.
Frequent Pattern Mining (Frequent Pattern Mining, FPM) is a kind of technology that data pattern is excavated, and is
Cluster, classification and correlation rule are laid a good foundation, and the application systems such as personalized web site, recommender system are widely used in
In.Wherein, the information of frequent mode and frequency include the privacy-sensitive data of user.
Guard method to privacy includes method for secret protection and difference method for secret protection etc. based on K- anonymity,
In, difference privacy is a kind of effectively to issue frequent mode and its hide user privacy information while frequency.Difference privacy
(Differential privacy, DP) [7] are the privacy leakage problems for being directed to staqtistical data base by Dwork [8] in 2006
The new secret protection model of the one kind proposed.Under this new definition, to the processing calculated result phase of database data set
Be for some variation specifically recorded it is insensitive, a certain item is individually recorded in data set or not in data set, to calculating
As a result influence almost can be ignored.Therefore, some data record because its be added data set in caused by privacy let out
Divulging a secret can nearly be limited in an acceptable, minimum range, so that attacker has no idea to pass through analytical calculation
As a result accurate individual record information is obtained.
Based on difference secret protection technology, it is hidden that energy effective protection hides user while issuing frequent mode and its frequency
Personal letter breath, so that risk of the sensitive information of user from disclosure.Mining Algorithms of Frequent Patterns tool with difference secret protection
There is important practical application meaning.
Summary of the invention
The object of the present invention is to provide a kind of Frequent Pattern Mining methods of efficient difference secret protection, solve existing
The serious problem of information leakage during Frequent Pattern Mining in the application of big data present in technology.
The technical scheme adopted by the invention is that a kind of Frequent Pattern Mining method of efficient difference secret protection, tool
Body follows the steps below to implement:
Step 1, input transaction data set (TDS) D, privacy budget ε, frequent mode number k, minimum support threshold value min_sup,
Privacy budget ε is split as ε1、ε2、ε3, wherein ε1Be using index mechanism pick out distributed when top-k frequent item set it is hidden
Private budget, ε2It is the privacy budget that Laplacian noise distribution is added for selected k frequent item set out, ε3It is selected for length
The privacy budget of Laplacian noise is added when selecting, and meets ε=ε1+ε2+ε3;Set Cset, Sset, Pset are emptied respectively,
Wherein Cset is candidate frequent item set, and the set that the top-k frequent mode that Sset index mechanism is picked out is formed, Pset is to add
Add the frequent mode set after noise;
Step 2 solves optimal transaction length lopt, and truncation, l before only retaining are carried out to transaction data set (TDS) Dopt?;
Step 3 excavates the fuzzy frequent itemsets that all support countings are not less than min_sup using FP-Growth method
Close Cset;
Step 4 carries out scale compression to the frequent mode set Cset that excavates using closing frequent mode;
Step 5 picks out top-k frequent mode using index mechanism and the true support counting of corresponding modes is formed
Set Sset, each mode p in the set sets up following formula:
Pr(p)∝exp(ε1× Rank (D, p)/2k),
Wherein, Rank (D, p) is the marking value of mode p,Wherein, as p ∈ ti
When, c (ti, p)=1;WhenWhen, c (ti, p)=0;
Step 6 adds Lap (k/ ε for the support counting of selected k mode out2) noise, form Pset;
Step 7 carries out consistency constraint processing to the mode support counting in Pset containing noise, and Lifting scheme is available
Property;
Step 8, output top-k frequent mode and noise count set RC.
The features of the present invention also characterized in that
Step 2 is specifically implemented according to the following steps:
Step 2.1, input raw data set D, privacy budget ε3, solve optimization length lopt;
Step 2.2, by conventions data collectionIt is set as empty;
Step 2.3 records r, l before truncation r retains for each in raw data set DoptItem is simultaneously added toIn.
Step 2.1 solves optimization length loptIt is specifically implemented according to the following steps:
Step 2.1.1, Z=< z is set1,z2,…,zi,…,z|D|>, ziFor the length value of i-th record in D;R=< rank
(z1),rank(z2),…,rank(zi),…,rank(z|D|) >, rank (zi) it is scoring functions;
Step 2.1.2, each length z is calculatediWeight W:
W=< exp (ε3×rank(z1)/2),exp(ε3×rank(z2)/2),…,exp(ε3×rank(z|D|)/2)>;
Step 2.1.3, weight in W is arranged to obtain orderly record dictionary according to descending;
Step 2.1.4, computational length zjThe probability selected:
Step 2.1.5, z is selected from WjAs optimization length lopt。
Step 4 is specifically implemented according to the following steps:
The frequent mode set Cset that step 4.1, input are generated through FP-Growth algorithm in the step 3 is as candidate
Collect Cset;
Step 4.2 carries out descending sort according to the scale of set of patterns to the Candidate Set Cset of input, then from maximum item
Collection scale starts to judge, to obtain the efficiency for improving selection candidate;
Step 4.3, to each of set Cset after sequence mode Cseti, i=1 ..., n judged, if
Mode CsetiIncluded in mode Csetj, in j=1 ..., n, and CsetiWith mode CsetjSupport counting it is equal, that is, meet
The definition for closing frequent mode, then illustrate CsetiThere are true hyper modes, set CsetiPosition is sky;Add CsetjTo set
SetAs obtain scale compression and simultaneously include full candidate collection information new set.
The beneficial effects of the invention are as follows a kind of Frequent Pattern Mining methods of efficient difference secret protection, as far as possible
Truncation is carried out to original transaction database in the case where more reservation effective informations, obtains the raising in efficiency;Another party
Face, after all frequent mode Candidate Sets are excavated in FP-Growth algorithm, frequent mode concept is closed in introducing, about subtracts Candidate Set
Scale.By two above step, so that the top-k Frequent Pattern Mining method of difference secret protection is imitated with preferable algorithm
Rate, especially in the transaction database that and transaction data set (TDS) more comprising things number is gradually increased, this method can mentioned
Frequent Pattern Mining is efficiently carried out while for secret protection.
Detailed description of the invention
Fig. 1 is a kind of Frequent Pattern Mining method of efficient difference secret protection of the present invention in pumsb-star data set
Upper excavation accuracy and general difference secret protection top-k Frequent Pattern Mining method excavate accuracy comparison diagram;
Fig. 2 is a kind of Frequent Pattern Mining method of efficient difference secret protection of the present invention in kosarak data set
On excavation accuracy and general difference secret protection top-k Frequent Pattern Mining method excavate accuracy comparison diagram;
Fig. 3 is that a kind of Frequent Pattern Mining method of efficient difference secret protection of the present invention and general difference privacy are protected
Protect top-k Frequent Pattern Mining method in frequent mode with data scale variation comparison diagram;
Fig. 4 is that a kind of Frequent Pattern Mining method of efficient difference secret protection of the present invention and general difference privacy are protected
Protect top-k Frequent Pattern Mining method in the time with data scale variation comparison diagram;
Fig. 5 is that a kind of Frequent Pattern Mining method of efficient difference secret protection of the present invention and general difference privacy are protected
Protect top-k Frequent Pattern Mining method in frequent item set scale with min_sup variation comparison diagram;
Fig. 6 is that a kind of Frequent Pattern Mining method of efficient difference secret protection of the present invention and general difference privacy are protected
Protect top-k Frequent Pattern Mining method in the time with min_sup variation comparison diagram.
Specific embodiment
The following describes the present invention in detail with reference to the accompanying drawings and specific embodiments.
A kind of Frequent Pattern Mining method of efficient difference secret protection of the present invention, is specifically implemented according to the following steps:
Step 1, input transaction data set (TDS) D, privacy budget ε, frequent mode number k, minimum support threshold value min_sup,
Privacy budget ε is split as ε1、ε2、ε3, wherein ε1Be using index mechanism pick out distributed when top-k frequent item set it is hidden
Private budget, ε2It is the privacy budget that Laplacian noise distribution is added for selected k frequent item set out, ε3It is selected for length
The privacy budget of Laplacian noise is added when selecting, and meets ε=ε1+ε2+ε3;Set Cset, Sset, Pset are emptied respectively,
Wherein Cset is candidate frequent item set, and the set that the top-k frequent mode that Sset index mechanism is picked out is formed, Pset is to add
Add the frequent mode set after noise;
Step 2 solves optimal transaction length lopt, and truncation, l before only retaining are carried out to transaction data set (TDS) Dopt,
It is specifically implemented according to the following steps:
Step 2.1, input raw data set D, privacy budget ε3, solve optimization length lopt;
Step 2.2, by conventions data collectionIt is set as empty;
Step 2.3 records r, l before truncation r retains for each in raw data set DoptItem is simultaneously added toIn.
Step 2.1 solves optimization length loptIt is specifically implemented according to the following steps:
Step 2.1.1, Z=< z is set1,z2,…,zi,…,z|D|>, ziFor the length value of i-th record in D;R=< rank
(z1),rank(z2),…,rank(zi),…,rank(z|D|) >, rank (zi) it is scoring functions;
Step 2.1.2, each length z is calculatediWeight W:
W=< exp (ε3×rank(z1)/2),exp(ε3×rank(z2)/2),…,exp(ε3×rank(z|D|)/2)>;
Step 2.1.3, weight in W is arranged to obtain orderly record dictionary according to descending;
Step 2.1.4, computational length zjThe probability selected:
Step 2.1.5, z is selected from WjAs optimization length lopt。
Step 3 excavates the fuzzy frequent itemsets that all support countings are not less than min_sup using FP-Growth method
Close Cset;
Step 4 carries out scale compression to the frequent mode set Cset that excavates using closing frequent mode, specifically according to
Lower step is implemented:
The frequent mode set Cset that step 4.1, input are generated through FP-Growth algorithm in the step 3 is as candidate
Collect Cset;
Step 4.2 carries out descending sort according to the scale of set of patterns to the Candidate Set Cset of input, then from maximum item
Collection scale starts to judge, to obtain the efficiency for improving selection candidate;
Step 4.3, to each of set Cset after sequence mode Cseti, i=1 ..., n judged, if
Mode CsetiIncluded in mode Csetj, in j=1 ..., n, and CsetiWith mode CsetjSupport counting it is equal, that is, meet
The definition for closing frequent mode, then illustrate CsetiThere are true hyper modes, set CsetiPosition is sky;Add CsetjTo set
SetAs obtain scale compression and simultaneously include full candidate collection information new set.
Step 5 picks out top-k frequent mode using index mechanism and the true support counting of corresponding modes is formed
Set Sset, each mode p in the set sets up following formula:
Pr(p)∝exp(ε1× Rank (D, p)/2k),
Wherein, Rank (D, p) is the marking value of mode p,Wherein, as p ∈ ti
When, c (ti, p)=1;WhenWhen, c (ti, p)=0;
Step 6 adds Lap (k/ ε for the support counting of selected k mode out2) noise, form Pset;
Step 7 carries out consistency constraint processing to the mode support counting in Pset containing noise, and Lifting scheme is available
Property;
Step 8, output top-k frequent mode and noise count set RC.
For primal algorithm DP-topkP and innovatory algorithm DP-OPtopkP, the experiment carried out includes the availability of algorithm
Experiment and efficiency experiment, data record used by wherein first part tests come fromhttp://fimi.ua.ac.be/ data/, wherein data set pumsb-star mainly describes census record data, and data set kosarak is mainly described
Web click steam records data, and the feature description of two datasets is as shown in table 1;Second part experimental data is used by program
The simulation collection data set automatically generated, the maximum length l=40 of fixed data set record, minimum length s=2, things item number
, by the scale of delta data collection, DP-topkP algorithm and DP- are observed in first group of group experiment of efficiency test for n=20
The variation of OPtopkP algorithm;In second group of experiment of efficiency test, simulated data sets are still used, fixed data set is passed through
Scale, change minimum threshold min_sup to observe the variation of DP-topkP algorithm and DP-OPtopkP algorithm.Experimental data set
Such as table 1:
The description of 1 test data set of table
(1) the usability testing process and result of secret protection mining algorithm
Testing standard in terms of the Result availability of algorithm is the average relative error according to algorithm, and the standard is main
It is that error brought by top-k frequent item set, formula are as follows in measurement publication D:
Wherein,Intermediate scheme piNoise support counting, TC (pi, TPk(D) indicate that it is true
Support counting.The smaller availability for illustrating algorithm of ARE value is higher.
It is obtained by testing with the ARE in kosarak data in pumsb-star data, i.e., algorithm operation is average opposite
Error, as depicted in figs. 1 and 2.
X-axis indicates that the k value of top-k, y-axis indicate obtained average relative error after algorithm operation in Fig. 1 and Fig. 2
ARE, this two width figure are the postrun average relative error comparison chart of algorithm.By ARE known in figure for DP-topkP algorithm and
DP-OPtopkP algorithm all keeps relative stability.It is very big that the ARE value of two kinds of algorithms differs not under same group of experiment.Illustrate this
The algorithm DP-OPtopkP of invention maintains the certain availability of output result under the premise of improved in terms of efficiency.
(2) algorithm operational efficiency is tested
1) algorithm changes in simulated data sets with data set scale
Experiment use simulated data sets, set ε=0.6, the k=100 of min_sup=10, tap-k, data set scale by
6KB to 96KB variation.Obtained result difference is as shown in Figure 3, Figure 4.
X-axis indicates the value of data set scale in Fig. 3, Fig. 4.Wherein y-axis indicates gained when two kinds of algorithm operations in Fig. 3
The candidate frequent item set scale arrived.From the analysis of the data of the figure it is found that DP-OPtopkP algorithm and DP-topkP algorithm are with number
According to the expansion of collection scale, obtained candidate's frequent mode scale is also in increased trend, and obtained by DP-OPtopkP algorithm
Candidate scale be less than DP-topkP algorithm, reason is the on the one hand preposition processing to database, on the other hand for
The compression of candidate scale.Y-axis indicates consumed time when two kinds of algorithm operations in Fig. 4, as can be seen from the figure DP-
The OPtopkP algorithm operation spent time runs the spent time less than DP-topkP algorithm, this is because reducing time
The raising of selected works scale and bring operational efficiency.
2) algorithm is compared in simulated data sets with the result that minimum threshold changes
The experiment of this group also uses simulated data sets, sets ε=0.6, the k=100 of top-k, and data set scale size is
128KB, changes the value of minimum threshold min_sup, and experimental result difference is as shown in Figure 5, Figure 6.
When minimum support threshold value min_sup value range is at [150,450], Fig. 5 gives DP-OPtopkP algorithm
With the obtained candidate frequent mode scale of DP-topkP algorithm, the obtained candidate frequent mode scale of the two algorithms is almost
It is close consistent, this is because the length of obtained Candidate Set scale and Candidate Set has when minimum threshold min_sup is larger
Limit, therefore there is no excessive shadow is generated to Candidate Set scale by the process of truncation transaction database and closed frequent item-sets compression
It rings, otherwise consumes a little time in break-in operation and squeeze operation.
As minimum threshold min_sup≤150, from fig. 5, it can be seen that DP-OPtopkP algorithm is obtained candidate frequent
Schema size is less than the obtained candidate frequent mode scale of DP-topkP algorithm, while comparison diagram 6, DP-OPtopkP algorithm
The spent time is also less than the time spent by DP-topkP algorithm.
The present invention reaches the optimization to DP-topkP algorithm by truncation item data library and reduction Candidate Set scale.Newly
Algorithm has the table of preferable aspect of performance on the transaction database that and transaction data set (TDS) more comprising things number is gradually increased
Existing, in terms of the availability of Result, innovatory algorithm maintains and maintains comparable level with reference to algorithm.
Claims (4)
1. a kind of Frequent Pattern Mining method of efficient difference secret protection, which is characterized in that specifically real according to the following steps
It applies:
Step 1, input transaction data set (TDS) D, privacy budget ε, frequent mode number k, minimum support threshold value min_sup, will be hidden
Private budget ε is split as ε1、ε2、ε3, wherein ε1It is that pick out the privacy distributed when top-k frequent item set pre- using index mechanism
It calculates, ε2It is the privacy budget that Laplacian noise distribution is added for selected k frequent item set out, ε3When for length selection
The privacy budget of Laplacian noise is added, and meets ε=ε1+ε2+ε3;Set Cset, Sset, Pset are emptied respectively, wherein
Cset is candidate frequent item set, and the set that the top-k frequent mode that Sset index mechanism is picked out is formed, Pset is that addition is made an uproar
Frequent mode set after sound;
Step 2 solves optimal transaction length lopt, and truncation, l before only retaining are carried out to transaction data set (TDS) Dopt?;
Step 3 excavates the frequent mode set that all support countings are not less than min_sup using FP-Growth method
Cset;
Step 4 carries out scale compression to the frequent mode set Cset that excavates using closing frequent mode;
The collection that step 5, the true support counting that top-k frequent mode and corresponding modes are picked out using index mechanism are formed
Sset is closed, each mode p in the set sets up following formula:
Pr(p)∝exp(ε1× Rank (D, p)/2k),
Wherein, Rank (D, p) is the marking value of mode p,Wherein, as p ∈ tiWhen, c
(ti, p)=1;WhenWhen, c (ti, p)=0;
Step 6 adds Lap (k/ ε for the support counting of selected k mode out2) noise, form Pset;
Step 7 carries out consistency constraint processing, Lifting scheme availability to the mode support counting in Pset containing noise;
Step 8, output top-k frequent mode and noise count set RC.
2. a kind of Frequent Pattern Mining method of efficient difference secret protection according to claim 1, which is characterized in that
The step 2 is specifically implemented according to the following steps:
Step 2.1, input raw data set D, privacy budget ε3, solve optimization length lopt;
Step 2.2, by conventions data collectionIt is set as empty;
Step 2.3 records r, l before truncation r retains for each in raw data set DoptItem is simultaneously added toIn.
3. a kind of Frequent Pattern Mining method of efficient difference secret protection according to claim 2, which is characterized in that
The step 2.1 solves optimization length loptIt is specifically implemented according to the following steps:
Step 2.1.1, Z=< z is set1,z2,…,zi,…,z|D|>, ziFor the length value of i-th record in D;R=< rank (z1),
rank(z2),…,rank(zi),…,rank(z|D|) >, rank (zi) it is scoring functions;
Step 2.1.2, each length z is calculatediWeight W:
W=< exp (ε3×rank(z1)/2),exp(ε3×rank(z2)/2),…,exp(ε3×rank(z|D|)/2)>;
Step 2.1.3, weight in W is arranged to obtain orderly record dictionary according to descending;
Step 2.1.4, computational length zjThe probability selected:
Step 2.1.5, z is selected from WjAs optimization length lopt。
4. a kind of Frequent Pattern Mining method of efficient difference secret protection according to claim 1, which is characterized in that
The step 4 is specifically implemented according to the following steps:
The frequent mode set Cset that step 4.1, input are generated through FP-Growth algorithm in the step 3 is as Candidate Set
Cset;
Step 4.2 carries out descending sort according to the scale of set of patterns to the Candidate Set Cset of input, then advises from maximum item collection
Mould starts to judge, to obtain the efficiency for improving selection candidate;
Step 4.3, to each of set Cset after sequence mode Cseti, i=1 ..., n are judged, if mode
CsetiIncluded in mode Csetj, in j=1 ..., n, and CsetiWith mode CsetjSupport counting it is equal, that is, meet close frequency
The definition of numerous mode, then illustrate CsetiThere are true hyper modes, set CsetiPosition is sky;Add CsetjTo setSetAs obtain scale compression and simultaneously include full candidate collection information new set.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910363516.5A CN110096900A (en) | 2019-04-30 | 2019-04-30 | A kind of Frequent Pattern Mining method of efficient difference secret protection |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910363516.5A CN110096900A (en) | 2019-04-30 | 2019-04-30 | A kind of Frequent Pattern Mining method of efficient difference secret protection |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110096900A true CN110096900A (en) | 2019-08-06 |
Family
ID=67446774
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910363516.5A Pending CN110096900A (en) | 2019-04-30 | 2019-04-30 | A kind of Frequent Pattern Mining method of efficient difference secret protection |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110096900A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110471957A (en) * | 2019-08-16 | 2019-11-19 | 安徽大学 | Localization difference secret protection Mining Frequent Itemsets based on frequent pattern tree (fp tree) |
CN112464277A (en) * | 2020-11-20 | 2021-03-09 | 东南大学 | Uncertain data privacy protection frequent item set publishing method |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109409128A (en) * | 2018-10-30 | 2019-03-01 | 南京邮电大学 | A kind of Mining Frequent Itemsets towards difference secret protection |
-
2019
- 2019-04-30 CN CN201910363516.5A patent/CN110096900A/en active Pending
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109409128A (en) * | 2018-10-30 | 2019-03-01 | 南京邮电大学 | A kind of Mining Frequent Itemsets towards difference secret protection |
Non-Patent Citations (4)
Title |
---|
卢国庆 等: "差分隐私下的一种频繁序列模式挖掘方法", 《计算机研究与发展》 * |
孙俊: "关联规则挖掘算法研究与改进", 《中国优秀硕士学位论文全文数据库(信息科技辑)》 * |
张啸剑 等: "差分隐私保护下一种精确挖掘top-k频繁模式方法", 《计算机研究与发展》 * |
赖娟等: "一种改进的压缩频繁模式挖掘算法", 《西南师范大学学报(自然科学版)》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110471957A (en) * | 2019-08-16 | 2019-11-19 | 安徽大学 | Localization difference secret protection Mining Frequent Itemsets based on frequent pattern tree (fp tree) |
CN110471957B (en) * | 2019-08-16 | 2021-10-26 | 安徽大学 | Localized differential privacy protection frequent item set mining method based on frequent pattern tree |
CN112464277A (en) * | 2020-11-20 | 2021-03-09 | 东南大学 | Uncertain data privacy protection frequent item set publishing method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Himberg et al. | Time series segmentation for context recognition in mobile devices | |
CN108595667A (en) | A kind of correlation analysis method of Network Abnormal data | |
CN109858930B (en) | Online transaction fraud detection method based on association graph characterization learning | |
JP2009099124A (en) | Method and system for data construction | |
Wang et al. | Automating transfer function design with valley cell‐based clustering of 2D density plots | |
CN110096900A (en) | A kind of Frequent Pattern Mining method of efficient difference secret protection | |
CN109284620A (en) | A kind of generation method, device and server for issuing data | |
CN103838754A (en) | Information searching device and method | |
CN109376231A (en) | A kind of media hotspot tracking and system | |
Sui et al. | Dynamic clustering scheme for evolving data streams based on improved STRAP | |
CN107506622B (en) | Software dynamic birthmark and plagiarism detection method based on memory object access sequence | |
Yin et al. | Nonnegative matrix factorization with bounded total variational regularization for face recognition | |
CN117290888B (en) | Information desensitization method for big data, storage medium and server | |
Skoumas et al. | On quantifying qualitative geospatial data: A probabilistic approach | |
CN113158206A (en) | Document security level dividing method based on decision tree | |
CN109254320A (en) | Seismic properties optimization and sand body Overlay District prediction technique based on image processing method | |
CN108170837A (en) | Method of Data Discretization, device, computer equipment and storage medium | |
Jiang | Research on Machine Learning Algorithm for Internet of Things Information Security Management System Research and Implementation | |
CN103902627A (en) | Data release method and device | |
Zhou et al. | Frequent sequence pattern mining with differential privacy | |
CN107529029A (en) | One kind adds tagged method, equipment and storage medium in file is detected | |
CN114240344A (en) | Enterprise personnel data processing method and device, computer equipment and storage medium | |
CN106778346A (en) | The method for secret protection of affairs type data flow issue | |
Abdulhak et al. | A systematic review of features identification and extraction for behavioral biometrie authentication in touchscreen mobile devices | |
CN110222528B (en) | Differential privacy protection method for frequent plot mining in data stream |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190806 |
|
RJ01 | Rejection of invention patent application after publication |