Mining method and system for privacy protection periodic high-utility item set

Abstract

The invention discloses a mining method and system for a privacy protection periodic high-utility item set, relates to the crossing field of data mining and privacy protection, mainly relates to the field of privacy protection utility mining, and comprises a selection stage and a hiding stage of a sensitive periodic high-utility item set, mining a periodic high-utility item set and obtaining a sensitive item set; the hiding stage comprises a preprocessing stage and a formal hiding stage; in the preprocessing stage, a sensitive item set list and a sensitive item list are constructed to serve as a list structure of a sensitive periodic high-utility item set and are stored in a database; and in a formal hiding stage, sensitive transactions and sensitive items in the database are selected and processed, so that the sensitive cycle high-utility item set cannot be mined in the database. According to the invention, sensitive information cannot be leaked when a user shares data; and the hidden database can be ensured not to influence the user to analyze the data.

Description

Technical Field

[0001] The present invention relates to the cross - field of data mining and privacy protection, mainly to the field of privacy - protected utility mining. Specifically, it relates to a method and system for mining privacy - protected periodic high - utility itemsets. Background Art

[0002] With the rapid development of the Internet, people have generated a large amount of data in the network. Although a large amount of data has been generated, most of the data is of no value. Therefore, how to discover valuable information from a large amount of data is crucial. The emergence of data mining technology effectively solves this problem, and it can mine meaningful knowledge from a large amount of data. As a branch of data mining, high - utility pattern mining can mine information with relatively high value from a large amount of data. However, the results generated by high - utility mining may have relatively high value in the short term, but they may not have relatively high value over time. For example, some products take advantage of customers' curiosity and have high sales in the short term but low sales in the long term. The proposed method for mining periodic high - utility itemsets can effectively mine the itemsets that have high utility in the database in the long term.

[0003] Although data mining technology can discover useful information from a large amount of data, some lawbreakers can also mine some sensitive information from the database. Regarding the problem of privacy leakage, when existing privacy - protected utility mining methods hide sensitive itemsets, they only consider the utility values of the itemsets and do not consider the period of the itemsets. Therefore, if existing privacy - protected utility mining algorithms are used to solve the problem of mining privacy - protected periodic high - utility itemsets, it will generate artificial cost (AC), which will interfere with users' decisions. Summary of the Invention

[0004] In order to solve the problem that the existing technology does not consider the period information of itemsets and thus cannot be effectively used to solve the problem of mining privacy - protected periodic high - utility itemsets, the purpose of the present invention is to provide a mining technology for privacy - protected periodic high - utility itemsets, aiming to solve the privacy - protection problem in the process of mining periodic high - utility itemsets.

[0005] To achieve the above - mentioned technical purpose, the present application provides a method for mining privacy - protected periodic high - utility itemsets, including: a selection stage and a hiding stage of sensitive periodic high - utility itemsets, where,

[0006] In the selection stage, mine periodic high - utility itemsets and obtain sensitive itemsets;

[0007] In the hiding stage, it includes a pre - processing stage and an official hiding stage;

[0008] In the preprocessing stage, a list of sensitive item sets and a list of sensitive items are constructed as the list structure of sensitive periodic high-utility item sets and stored in the database;

[0009] In the formal hiding stage, the sensitive transactions and sensitive items in the database are selected and processed so that sensitive periodic high-utility item sets cannot be mined from the database.

[0010] Preferably, in the selection stage, the PHM algorithm is used to mine all eligible periodic high-utility item sets, and a random algorithm is used to select sensitive item sets from the periodic high-utility item sets.

[0011] Preferably, in the preprocessing stage, the list of sensitive item sets is a five-tuple structure, expressed as: [SIS, SU, <tid:utility>, LP, sup], where SIS represents S, and S is a sensitive periodic high-utility itemset; SU represents the utility value of S; <tid:utility>It represents the transaction identifier of the transaction containing S and the utility value in this transaction; LP represents the maximum period of S; sup represents the support degree of S, and the average period of the item set is calculated using sup to determine whether an item set meets the periodic constraint.

[0012] Preferably, in the preprocessing stage, the sensitive item list is a five-tuple with the structure: [tid, si, iu, cnt, mp], where tid represents the transaction identifier of the transaction containing S; si represents the sensitive item in S; iu represents the utility value of this sensitive item in S; cnt represents the quantity of the sensitive item in S; mp represents the maximum period of the sensitive item.

[0013] Preferably, in the formal hiding stage, when processing sensitive transactions, the transaction that maximizes the utility value of the sensitive period high-utility item set is selected for priority processing.

[0014] Preferably, in the formal hiding stage, when processing sensitive items, according to the maximum period of the sensitive items, the items with the maximum and minimum maximum periods are respectively selected for priority processing.

[0015] Preferably, in the formal hiding stage, the sensitive items in the sensitive transaction are deleted or the quantity of the sensitive items in the sensitive transaction is modified, so that the utility or period of the sensitive item set does not meet its threshold constraint, achieving the hiding effect.

[0016] Preferably, in the preprocessing stage, based on the constructed sensitive item set list and sensitive item list, the sensitive item sets are sorted in descending order according to the number of items contained in the sensitive item sets to form an ordered list structure of sensitive periodic high-utility item sets.

[0017] Preferably, when sorting the sensitive item sets in descending order, the minimum support threshold is calculated and the minimum support threshold is applied to the formal hiding stage to complete the hiding of the sensitive periodic high-utility item sets, where the minimum support threshold minSup = |D| / maxAvg - 1, in the formula, |D| is the size of the database D, and maxAvg represents the maximum average period threshold.

[0018] The present invention also discloses a mining system for privacy-preserving periodic utility, which is used to implement a mining method for privacy-preserving periodic high-utility item sets, including:

[0019] A mining module, which is used to mine periodic high-utility item sets and obtain sensitive item sets;

[0020] A hidden module, which is used to construct a list of sensitive item sets and a list of sensitive items, store them in the database as a list structure of sensitive periodic high-utility item sets; meanwhile, select and process sensitive transactions and sensitive items in the database so that sensitive periodic high-utility item sets cannot be mined from the database.

[0021] The present invention discloses the following technical effects:

[0022] 1. The present invention can ensure that users will not disclose sensitive information when sharing data;

[0023] 2. The present invention can ensure that the hidden database does not affect users' analysis of data;

[0024] 3. The present invention provides a secure mining and analysis method for periodic high-utility item sets. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0026] Figure 1 is the model diagram of the present invention;

[0027] Figure 2 is the workflow diagram of the present invention;

[0028] Figure 3 is the algorithm pseudo code of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0029] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, rather than all embodiments. Usually, the components of the embodiments of the present application described and illustrated in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the present application claimed, but only represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0030] Such as Figure 1 As shown in the figure, the present invention provides a mining technology for privacy - protected periodic high - utility itemsets, including: a selection stage and a hiding stage of sensitive periodic high - utility itemsets, specifically including the following:

[0031] The selection stage includes a mining stage of periodic high - utility itemsets and the discovery of sensitive periodic high - utility itemsets; in the mining stage of periodic high - utility itemsets, all eligible periodic high - utility itemsets are mined using the existing PHM algorithm; in the discovery stage of sensitive periodic high - utility itemsets, a specified number of sensitive itemsets are randomly selected from the mined periodic high - utility itemsets using a random algorithm.

[0032] The hiding stage includes a pre - processing stage and an official hiding stage of sensitive periodic high - utility itemsets;

[0033] In the pre - processing stage, a list structure of sensitive periodic high - utility itemsets is constructed and the sensitive periodic high - utility itemsets are sorted. Among them, the original database is modified using a hiding algorithm to obtain a cleaned database; no sensitive periodic high - utility itemsets can be mined from the cleaned database, and at this time the data owner can safely share the data;

[0034] Constructing the list structure of sensitive periodic high - utility itemsets is to construct a sensitive itemset list (SISL) and a sensitive item list (SIL) for each sensitive itemset. Among them, the SISL structure is a five - tuple, expressed as: [SIS, SU, <tid:utility>, LP, sup]. Given a sensitive periodic high-utility itemset S, SIS represents S; SU represents the utility value of S; <tid:utility>The transaction identifier of the transaction in which S appears and the utility value in that transaction; LP represents the maximum period of S; sup represents the support degree of S. The average period of the item set can be calculated using sup to determine whether an item set meets the periodic constraint. The SIL structure is also a five-tuple, and its structure is: [tid, si, iu, cnt, mp]. Among them, tid represents the transaction identifier of the transaction containing S; si represents the sensitive item in S; iu represents the utility value of the sensitive item in S; cnt represents the number of sensitive items in S; mp represents the maximum period of the sensitive item.

[0035] Using these data structures can save the information of the sensitive item set in the database, thus avoiding multiple scans of the database;

[0036] The sorting of sensitive period high-utility item sets is the processing order of each sensitive item set in the hiding algorithm. The present invention preferentially selects the sensitive period high-utility item set with the largest number of items for processing. An appropriate processing order can speed up the execution of the algorithm.

[0037] The formal hiding phase includes the selection of sensitive transactions, the selection of sensitive items, and the processing of sensitive items:

[0038] The selection of sensitive transactions is the processing order of all transactions containing sensitive item sets. Different sensitive transaction selection strategies will affect the performance of the algorithm. The present invention preferentially selects the transaction that maximizes the utility value of the sensitive period high-utility item set for processing;

[0039] The selection of sensitive items is the processing order of the sensitive items in the sensitive item sets in the selected sensitive transactions, which is used to determine which sensitive item the algorithm preferentially processes. The present invention adopts different selection strategies according to the maximum period of the sensitive items: preferentially selects the items with the largest maximum period and the smallest maximum period for processing respectively;

[0040] The processing of sensitive items is to delete the sensitive items in the sensitive transactions or modify the quantity of the sensitive items in the sensitive transactions, so that the utility or period of the sensitive item set does not meet its threshold constraint, achieving the hiding effect.

[0041] As Figure 2 shown, the mining method provided by the present invention specifically includes the following steps:

[0042] Step 1: Use the periodic high-utility item set mining algorithm to mine all high-utility item sets that meet the periodic constraint;

[0043] Step 2: Select the sensitive item sets to be hidden from the mined periodic high-utility item sets;

[0044] Step 3: Use the privacy-preserving periodic high-utility item set mining algorithm to hide the selected sensitive item sets.

[0045] Specifically, in step 1, select the sensitive period high-utility item set that you want to hide;

[0046] Step 1.1: Use the existing PHM algorithm to mine all eligible periodic high-utility item sets, assuming that the minimum utility threshold is α, the maximum period threshold is maxPer, the maximum average period threshold is maxAvg, the minimum period threshold is minPer, and the minimum average period threshold is minAvg. This threshold setting is adopted in the following sections of the present invention;

[0047] Step 1.2: Use a random algorithm to randomly select a specified number of sensitive periodic high-utility item sets (SPIs) from the periodic high-utility item sets mined above;

[0048] In step 2, the SPIs are pre-processed by recording the information of the SPIs in the database, thereby avoiding multiple database scans;

[0049] Step 2.1: For each sensitive period high utility item set (sp k ) construct its SISL and SIL structures;

[0050] Step 2.2: According to sp k The number of items contained in,sort the sensitive period high utility item sets in SPIs in descending order;

[0051] Step 2.3: Calculate the minimum support threshold minSup=|D| / maxAvg-1 according to maxAvg, where |D| is the size of database D.

[0052] In step 3, the SPIs are formally hidden;

[0053] Step 3.1: According to sp k SISL, obtain sp k The utility value of SISL(sp k ).SU、sp k The support degree SISL(sp k ).sup and spk maximum period SISL (sp k LP, thereby calculating the modified utility value du(sp k )=SISL(sp k ).SU-α, current support ds(sp k )=SISL(sp k ).sup, the current maximum cycle dp (sp k )=SISL(sp k ).LP.

[0054] Step 3.2: Determine whether du(sp k )≥0, dp(sp k )≤maxPer, and ds(sp k )≥minSup hold simultaneously. If so, it means that sp k is still a periodic high-utility itemset, and it is necessary to go to Step 3.3 to hide sp k . If not, it means that sp k is not a periodic high-utility itemset, the hiding is successful, and go to Step 3.9.

[0055] Step 3.3: Select the transaction with the maximum utility value as the victim transaction T k , where T vic = max{u(sp vic ,T k ),T d ),T d ∈SISL.[tid:utility]};

[0056] Step 3.4: If the item with the maximum period in the victim transaction T vic for sp k is selected as the victim item I vic , then I vic = max{mp(i k ,T vic ), 1≤k≤|sp k |, i k ∈SIL(sp k ,T vic )}, if the strategy adopted is to select the item with the minimum period in the victim transaction T vic for sp k as the victim item I vic , then I vic = min{mp(i k ,T vic ), 1≤k≤|sp k |, i k ∈SIL(sp k ,T vic )};

[0057] Step 3.5: According to the SIL of sp k , determine whether the utility value of I vic is greater than du(sp k ). If so, go to Step 3.6; if not, go to Step 3.7.

[0058] Step 3.6: Modify the quantity of the victim item I vic , and calculate the quantity of I that needs to be subtracted vic The quantity dq is du(sp k ) divided by I vic The external utility value eu(I vic ) rounded up. SIL(I vic , T vic ).cnt -= dq, SIL(I vic , T vic ).iu -= dq × eu(I vic ), where sp k no longer meets the minimum utility threshold and is thus removed from the SPIs k . Then, jump to Step 3.8.

[0059] Step 3.7: Remove I vic from T vic , and recalculate the utility value SISL(sp k ).su = u(I k , T vic ), the support SISL(sp vic ).sup--, and the maximum period SISL(sp k ).LP. Then, jump to Step 3.8. k ).LP. Then, jump to Step 3.8.

[0060] Step 3.8: Update the SPIs, SISL, and SIL, and jump to Step 3.2.

[0061] Step 3.9: Update the database.

[0062] As Figure 3 shown, the mining method provided by the present invention has its algorithm pseudocode expressed as follows:

[0063] This algorithm can clean the original database and successfully and effectively hide sensitive cyclic high-utility item sets. The algorithm can be divided into two parts: the preprocessing stage of sensitive cyclic high-utility item sets and the hiding stage. In the preprocessing stage, the SISL and SIL structures are used to accelerate the execution time of the algorithm and improve its efficiency.

[0064] The first stage is the preprocessing of the SPIs (Lines 1 - 5). First, construct the SISL and SIL structures for each sensitive item set in the SPIs (Line 2). Then, sort the sensitive item sets in descending order according to the number of items contained in the sensitive item sets (Line 4). In the algorithm, longer sensitive item sets are processed first. Calculate the minimum support threshold minSup according to maxAvg (Line 5).

[0065] The second stage is the hiding operation of SPIs (Lines 6 - 28). The algorithm iteratively processes each element sp in the SPIs k until all sensitive item sets are hidden. For each sensitive item set sp k , its SISL and SIL structures are obtained (Line 7). The threshold information of the item set is calculated based on its SISL, du(sp k ), ds(sp k ), dp(sp k ) (Lines 8 - 10). du(sp k ) represents the difference between the utility of sp k and α. ds(sp k ) represents the support of sp k . dp(sp k ) represents the maximum period of sp k . Based on the SISL of sp k , the transaction with the maximum utility value of sp k is selected as the victim transaction for modification first (Line 12). Then, according to the SIL of sp k , the item with the maximum maxPer or the minimum maxPer in sp k is selected as the victim item (Line 13). Whether to delete the item or modify the quantity of the item is determined according to the utility of the victim item in the victim transaction (Lines 14 - 27). If the utility of the victim item in the victim transaction is not greater than du(sp k ), the victim item is deleted from the victim transaction, and the SPIs, SISL, and SIL are updated (Lines 14 - 19). Otherwise, the quantity of the victim item in the victim transaction is reduced (Line 22). Then, sp k can be removed from the SPIs because it has been hidden (Line 24). Then the SPIs, SISL, and SIL are updated (Line 25). When all SPIs are hidden, the algorithm terminates.

[0066] In summary, the present invention proposes two novel data structures: Sensitive Itemset List (SISL) and Sensitive Item List (SIL); and, the sensitive item selection strategy adopted by the present invention, which selects the victim item according to the period of the sensitive item; compared with the existing technologies, the present invention can ensure that users will not disclose sensitive information when sharing data; the present invention can ensure that the hidden database does not affect users' analysis of data.

[0067] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processors of general-purpose computers, special-purpose computers, embedded processors, or other programmable data processing devices to generate a machine, such that the instructions executed by the processors of the computer or other programmable data processing devices produce means for implementing the functions specified in one Figure 1 flow or multiple flows and / or blocks Figure 1 or means for implementing the functions specified in one block or multiple blocks.

[0068] In the description of the present invention, it should be understood that the terms "first" and "second" are used for descriptive purposes only and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, "a plurality of" means two or more unless otherwise specifically defined.

[0069] Obviously, those skilled in the art can make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.< / tid:utility> < / tid:utility> < / tid:utility> < / tid:utility>

Claims

1. A method for mining privacy-preserving periodic high-utility itemsets, characterized in that: include: The selection and hiding stages of sensitive periodic high-utility itemsets, where In the selection stage, periodic high-utility item sets are mined and sensitive item sets are obtained; The hiding stage includes a pre-processing stage and a formal hiding stage; In the preprocessing stage, a sensitive item set list and a sensitive item list are constructed as a list structure of sensitive periodic high-utility item sets and stored in a database; In the formal hiding stage, sensitive transactions and sensitive items in the database are selected and processed so that sensitive period high-utility item sets cannot be mined in the database.

2. A method for mining privacy-preserving periodic high-utility itemsets according to claim 1, characterized in that: In the selection phase, the PHM algorithm mines all eligible periodic high-utility item sets and uses a random algorithm to select sensitive item sets from the periodic high-utility item sets.

3. A method for mining privacy-preserving periodic high-utility itemsets according to claim 2, characterized in that: In the preprocessing stage, the sensitive item set list is a five-tuple structure, represented as: [SIS, SU, <tid:utility>,LP,sup], where SIS represents S, S is the sensitive periodic high utility item set; SU represents the utility value of S; <tid:utility> It represents the transaction identifier of the transaction containing S and the utility value in the transaction; LP represents the maximum period of S; sup represents the support of S. sup is used to calculate the average period of the itemset to determine whether an itemset satisfies the periodicity constraint.< / tid:utility> < / tid:utility> 4. A method for mining privacy-preserving periodic high-utility itemsets according to claim 3, characterized in that: In the preprocessing stage, the sensitive item list is a quintuple with the structure [tid, si, iu, cnt, mp], where tid represents the transaction identifier of the transaction containing S; si represents the sensitive item in S; iu represents the utility value of the sensitive item in S; cnt represents the number of sensitive items in S; and mp represents the maximum period of the sensitive item.

5. A method for mining privacy-preserving periodic high-utility itemsets according to claim 4, characterized in that: In the formal hiding stage, when processing sensitive transactions, the transactions that maximize the utility value of the sensitive period high-utility item set are selected for priority processing.

6. A method for mining privacy-preserving periodic high-utility itemsets according to claim 5, characterized in that: In the formal hiding stage, when processing sensitive items, according to the maximum period of sensitive items, the items with the largest maximum period and the smallest maximum period are selected for priority processing.

7. A method for mining privacy-preserving periodic high-utility itemsets according to claim 6, characterized in that: In the formal hiding stage, the sensitive items in the sensitive transaction are deleted or the number of sensitive items in the sensitive transaction is modified, so that the utility or period of the sensitive item set does not meet its threshold constraint, thereby achieving the hiding effect.

8. A method for mining privacy-preserving periodic high-utility itemsets according to claim 7, characterized in that: In the preprocessing stage, based on the constructed sensitive item set list and sensitive item list, the sensitive item sets are sorted in descending order according to the number of items contained in the sensitive item sets to form an ordered list structure of sensitive periodic high-utility item sets.

9. A method for mining privacy-preserving periodic high-utility itemsets according to claim 8, characterized in that: When the sensitive item set is sorted in descending order, the minimum support threshold is calculated and applied to the formal hiding stage to complete the hiding of the sensitive periodic high-utility item set, where the minimum support threshold minSup = |D| / maxAvg-1, where |D| is the size of the database D and maxAvg represents the maximum average period threshold.

10. A system for mining privacy-preserving periodic high-utility itemsets, used to implement a method for mining privacy-preserving periodic high-utility itemsets as claimed in claim 1, characterized in that: include: Mining module, used to mine periodic high-utility itemsets and obtain sensitive itemsets; The hidden module is used to construct a sensitive item set list and a sensitive project list as a list structure of sensitive periodic high-utility item sets, and store them in a database; at the same time, sensitive transactions and sensitive projects in the database are selected and processed so that sensitive periodic high-utility item sets cannot be mined in the database.

Images