A method, system and device for mining periodic utility itemsets under incremental database

Abstract

The present invention provides a method, system and device for mining periodic utility item sets under an incremental database, including obtaining an original transaction data set; obtaining candidate items of the original transaction data set; obtaining an incremental utility list of the candidate items; screening the incremental utility list of the candidate items to obtain the periodic high-utility items of the original database; screening the candidate items to obtain the periodic high-utility item set of the original database; scanning the incremental part of the database to obtain the incremental transaction data set; screening the incremental transaction data set and screening out new candidate items from the candidate items; expanding the incremental utility list of the new candidate items and screening them to obtain the periodic high-utility items of the incremental database; the present invention realizes efficient and dynamic data mining through pruning strategies and dynamic maintenance of utility lists, ensuring performance and accuracy when processing large-scale and continuously updated data sets, and its dynamic adaptability supports real-time data changes and accurately identifies periodic patterns with high utility value.

Description

Technical Field

[0001] The present invention relates to the technical field of data mining, and in particular to a method, system and device for mining periodic utility item sets in an incremental database. Background Art

[0002] Data mining refers to the process of searching for hidden information from a large amount of data through algorithms. Incremental databases refer to database systems where data is continuously added or updated over time. Periodic utility mining is based on this, looking for information patterns that not only have periodic patterns but also have high utility value for the decision-making process.

[0003] The existing common data mining direction is the mining algorithm based on high-utility itemsets. This direction gives a single item a unit profit or weight, and pays more attention to the profit level between different items compared to the traditional frequent itemset mining algorithm. This direction mainly studies whether an item or item set (or pattern) has high utility. If a pattern is highly useful, it can be considered that the pattern conforms to the information pattern to be mined. However, this direction faces a significant challenge: the set of high-utility itemsets mined by the traditional high-utility itemset mining algorithm is too large, making it difficult for analysts to effectively extract usable information patterns.

[0004] In addition, this direction is not good at identifying repeated and regular purchasing behaviors, although such behaviors are very common in the real world. For example, in a retail store, certain combinations of products may be purchased together every day or every week. Accurately capturing this information can greatly enhance the understanding of customer purchasing behavior, thereby guiding purchasing and sales staff. For example, purchasing staff can make wise replenishment decisions based on these product combinations, while sales staff can design targeted promotions around them. Existing periodic utility item set mining algorithms include PHM, PHMN+, etc. These algorithms can be preset by users with minimum utility thresholds and maximum period thresholds to mine qualified periodic high-utility item sets.

[0005] The current problem is that as the amount of data continues to increase, it becomes a challenge to quickly and efficiently discover useful patterns in dynamically updated data sets. Summary of the invention

[0006] In view of the deficiencies of the prior art, the present invention provides a method, system and device for mining periodic utility item sets in an incremental database, which can mine high-profit and periodic item sets in a dynamically updated database and provide users with useful models.

[0007] The technical solution of the present invention is: a method for mining periodic utility item sets under an incremental database, comprising the following steps:

[0008] S1), scanning the transaction information of the original database to obtain the original transaction data set;

[0009] S2), using pruning strategy to filter the original transaction data set to obtain candidate items;

[0010] S3), scan the original transaction data set for the second time to obtain the incremental utility list of candidate items;

[0011] S4), screening the incremental utility list of candidate items based on a user-defined minimum utility threshold to obtain periodic high-utility items of the original database;

[0012] S5), using the pruning strategy to screen the candidate items, and performing a connection operation on the incremental utility lists of the screened candidate items to obtain a periodic high-utility item set of the original database;

[0013] S6), when a new transaction occurs, the database record increases, and the incremental part of the database is scanned to obtain the incremental transaction data set;

[0014] S7), screening the incremental transaction data set, and screening out new candidate items from the candidate items in step S2);

[0015] S8), scanning the incremental transaction data set for the second time to expand the incremental utility list of new candidates;

[0016] S9), screening the incremental utility list of new candidate items based on a minimum utility threshold defined by a user to obtain periodic high utility items of the incremental database;

[0017] S10), using the pruning strategy in step S5) to screen new candidate items, and performing a connection operation on the incremental utility lists of the screened new candidate items to obtain a periodic high-utility item set of the incremental database;

[0018] S11), for new transactions that are continuously added, steps S6) to S10) are periodically executed to obtain the latest periodic high-utility item set of the incremental database.

[0019] Preferably, in step S2), the original transaction data set is screened using a pruning strategy to obtain candidate items, which specifically includes the following steps:

[0020] S21), calculating the TWU value and maxper value of each item in the original transaction data set;

[0021] S22) Define the minimum utility threshold minutil and the maximum period threshold maxPer according to user needs, filter out candidate items that meet these two thresholds, and sort the candidate items in ascending order according to the size of the TWU value, and obtain the set of candidate items as I*. By filtering the item sets in the original transaction data set, remove all single items and their extended item sets that cannot become periodic high-utility item sets.

[0022] Preferably, step S3) specifically includes the following steps:

[0023] S31), scan the original transaction database again, and calculate the internal utility of all candidate items in the candidate item set I*;

[0024] S32), calculating the remaining utility of all candidate items;

[0025] S33), constructing an incremental utility list of candidate items;

[0026] S34), insert the incremental utility lists of all candidate items into the global list of single incremental utility lists (LUL), and sort them in the order of I*. The information saved in the single incremental utility list can be directly mined without scanning the original transaction data set again; the created global list is used to guide the connection order of subsequent incremental utility lists, as well as the incremental utility list reconstruction problem related to the incremental transaction after the incremental transaction is added to the database.

[0027] Preferably, step S4) specifically includes the following steps:

[0028] S41), calculating the utility value of the candidate item according to the information of the incremental utility list;

[0029] S42), based on the minimum utility threshold minutil defined by the user, compare the utility values ​​of the candidate items to obtain the periodic high utility items.

[0030] Preferably, step S5) specifically includes the following steps:

[0031] S51), calculating the total internal utility sumIutil plus the total residual utility sumRutil of the candidate items according to the information of the incremental utility list;

[0032] S52), based on the minimum utility threshold minutil defined by the user, compare the total internal utility sumIutil and the total residual utility sumRutil to obtain the screened candidate items;

[0033] S53), for the incremental utility lists corresponding to these screened candidate items, a depth-first traversal is performed to expand them, and a periodic high-utility item set of the original database is obtained.

[0034] Preferably, step S7) specifically includes the following steps:

[0035] S71), locate the incremental transaction part of the incremental transaction data set, and calculate the maxPer and TWU values ​​of each candidate item in the set I* of candidate items obtained in step S2);

[0036] S72), define the minimum utility threshold minutil and the maximum period threshold maxPer according to user needs, filter out new candidate items that meet these two thresholds, and sort the new candidate items in ascending order according to the size of the TWU value, and update the set I* of new candidate items.

[0037] Preferably, step S8) specifically includes the following steps:

[0038] S81), scanning the incremental transaction part of the incremental transaction data set again, and calculating the internal utility and remaining utility of the incremental transaction part corresponding to the candidate items in I*;

[0039] S82), expanding the list of incremental utilities of these candidates;

[0040] S83), deleting the incremental utility lists corresponding to the items not in I* from the global list (LUL) of the single incremental utility list, and sorting the LUL according to the order of I*;

[0041] S84) Update the remaining utility value of each incremental utility list in the LUL.

[0042] By reconstructing the incremental utility list of the new candidate single item, the saved information can be directly mined without scanning the incremental transaction data set again; the created global list is used to guide the connection order of subsequent incremental utility lists, as well as the reconstruction of related incremental utility lists after subsequent incremental transactions are added to the database.

[0043] Preferably, the present invention further provides a system for mining periodic utility itemsets in an incremental database, the system comprising:

[0044] A data acquisition module is used to scan the transaction information of the original database and the subsequent incremental database to obtain the original transaction data set or the incremental transaction data set;

[0045] The data screening module is used to screen the transaction data set to obtain candidate items;

[0046] The utility list construction and maintenance module is used to construct the utility list of the corresponding items for the candidate items, obtain the corresponding utility list, and maintain the incremental part;

[0047] The utility list merging module is used to merge the utility lists corresponding to single items or item sets that meet the pruning strategy, and determine whether they are periodic high-utility item sets.

[0048] Preferably, the present invention further provides a device for mining periodic utility item sets in an incremental database, comprising:

[0049] at least one processor;

[0050] at least one memory for storing at least one program;

[0051] When at least one of the programs is executed by at least one of the processors, the at least one processor implements the mining method.

[0052] The beneficial effects of the present invention are:

[0053] 1. The present invention realizes efficient and dynamic data mining through advanced pruning strategies and dynamic maintenance of utility lists, ensuring performance and accuracy when processing large-scale and constantly updated data sets. Its dynamic adaptability supports real-time data changes, accurately identifies periodic patterns with high utility value, and provides timely and accurate support for decision-making;

[0054] 2. The modular design and easy implementation and scalability of the hardware implementation scheme of the present invention further enhance its applicability and practical value in different data mining tasks, greatly improving business decision-making efficiency and performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is a flowchart of the method of Example 1 of the present invention;

[0056] Figure 2 Schematic diagram of the process of the method of Example 1 of the present invention;

[0057] Figure 3 This is a system framework diagram of Embodiment 3 of the present invention; DETAILED DESCRIPTION

[0058] The specific implementation of the present invention will be further described below in conjunction with the accompanying drawings:

[0059] Example 1

[0060] This embodiment provides a method for mining periodic utility itemsets in an incremental database. This embodiment describes the following concepts:

[0061] Transaction: A record in the transaction database.

[0062] Item: Each information item set recorded in a transaction, and a transaction contains at least one item.

[0063] Itemset: A set of at least one data item, used to represent an association rule within the database.

[0064] Utility: The utility or profit corresponding to each data item in the database is expressed as utility.

[0065] Itemset residual utility: In a specified transaction, the sum of the utilities of all items in the set that are sorted after the specified itemset, except for the itemset utility of the specified itemset.

[0066] Maximum period of an itemset: Assuming that an itemset appears in virtual transaction numbers 0 and n+1 (n is the number of the last transaction), the difference between the numbers of the two transactions with the largest interval between the itemset in the current data set.

[0067] like Figure 1 and 2 As shown, the method comprises the following steps:

[0068] S1) Scan the transaction information of the original database to obtain the original transaction data set; specifically:

[0069] Extract relevant information from the original database, and construct the original data set D according to the number of transaction items in the transaction and the unit profit of the items, where D = {T1, T2, ..., T n}, which means there are n transactions, and different transactions represent a transaction at a certain time, including the sale of several items. The specific utility calculation method for a single item is as follows:

[0070] U(i,T c )=q(i,T c )*p(i);

[0071] In the formula, U(i,T c ) indicates transaction T c The utility value of item set i, q(i,T c ) indicates transaction T c The number of item sets i in , p(i) represents the unit profit of item set i. This embodiment can calculate the total utility of item set i under the original data set D, and then determine whether item set i is a high-utility item set according to the minimum utility threshold minutil for determining whether it is a high-utility item set.

[0072] S2) Use the pruning strategy to filter the original transaction data set to obtain candidate items; specifically, the steps include:

[0073] S21), calculating the TWU value and maxper value of each item in the original transaction data set;

[0074]

[0075]

[0076] Where TWU(X) represents the sum of the utility values ​​of all transactions containing item set X, tu(T c ) represents the utility value of the transaction containing the item set X, u(i j ) shows transaction T c Middle Item j According to the definition of TWU, the upper bound of TWU has a downward closure property, that is, the TWU value of the expanded itemset of itemset X must not be greater than the TWU value of itemset X.

[0077] The maxper value of the itemset X is used as a guarantee that the itemset has periodic properties. Its calculation expression is as follows:

[0078] maxper(X)=max(ps(X));

[0079] ps(X)={g1-g0,g2-g1,...,g k+1 -g k}

[0080] In the formula, ps(X) item set X appears k times in the original transaction data set D, g1 to g k represents the order of the transaction in which the item set X appears in D, while g0 and g k+1 are 0 and |D|+1 respectively.

[0081] When the maxper value of item set X is not greater than the maximum period threshold maxPer, it means that item set X has periodicity.

[0082] S22), define the minimum utility threshold minutil and the maximum cycle threshold maxPer according to user needs, filter out candidate items that meet the two thresholds, and sort the candidate items in ascending order according to the size of the TWU value, and obtain a set of candidate items I*. By calculating the TWU value and maxper value of all items in the data set D, and then giving the minimum utility threshold minutil and the maximum cycle threshold maxPer, filter out candidate items that meet the two thresholds.

[0083] For the set of candidate items selected, sort them in ascending order according to their TWU values. The purpose is as follows: If the obtained sorting is {A < B < C < D < E}, then when expanding the item set according to the sorting subsequently, the same item set will not be generated repeatedly; and according to relevant proofs, the number of expansions required by this method of sorting in ascending order is the least, and finally the set of candidate items I* is obtained.

[0084] By screening the item sets in the original transaction data set, remove all single items that cannot become periodic high-utility item sets and their expanded item sets.

[0085] S3) Scan the original transaction data set for the second time to obtain the incremental utility list of candidate items; specifically as follows:

[0086] S31) Scan the original transaction database again to calculate the internal utility of all candidate items in the set of candidate items I*;

[0087] For the calculation of the internal utility, its expression is as follows:

[0088] iutil(X) = u(X,T tid )

[0089] In the formula, the internal utility iutil(X) represents the utility of the item set X in the transaction T tid ;

[0090] S32) Calculate the remaining utility of all candidate items; for the calculation of the remaining utility, its expression is as follows:

[0091]

[0092] The remaining utility rutil(X) represents the sum of the utilities of all items sorted after X in the transaction T tid ;

[0093] S33) Construct the incremental utility list of candidate items; the incremental utility list is a list, and each element of it is a tuple containing 3 elements: namely Tid, iutil, and rutil; used to save the detailed information of each item set.

[0094] S34) Insert the incremental utility lists of all candidates into the global list (LUL) of single-item incremental utility lists and sort them in the order of I*. The information stored in the single-item incremental utility list can be directly used for mining without scanning the original transaction dataset again; the created global list is used to guide the connection order of subsequent incremental utility lists and the reconstruction of relevant incremental utility lists after incremental transactions are added to the database.

[0095] S4) Screen the incremental utility lists of candidates based on a user-defined minimum utility threshold to obtain the periodic high-utility items in the original database. Specifically:

[0096] S41) Calculate the utility value of a candidate based on the information in the incremental utility list.

[0097] S42) Compare the utility value of a candidate with the user-defined minimum utility threshold minutil. If the utility value of this item is not less than the minimum utility threshold minutil, then this item is a periodic high-utility item, and the periodic high-utility items are obtained.

[0098] S5) Use a pruning strategy to screen candidates, and perform a join operation on the incremental utility lists of the screened candidates to obtain the periodic high-utility item set of the original database.

[0099] S51) Calculate the total internal utility sumIutil plus the remaining utility sumRutil of a candidate based on the information in the incremental utility list as a pre-computation for the subsequent pruning strategy.

[0100] S52) Compare the total internal utility sumIutil plus the remaining utility sumRutil with the user-defined minimum utility threshold minutil. If sumIutil(X)+sumRutil(X)<minutil, then prune this item set X and its extended item sets; otherwise, obtain the set of screened candidates.

[0101] S53) For the incremental utility lists corresponding to these screened candidates, perform a depth-first traversal for expansion to obtain the periodic high-utility item set of the original database.

[0102] S6) When new transactions occur and the database records increase, scan the incremental part of the database to obtain the incremental transaction dataset.

[0103] In this embodiment, relevant information is extracted from the incremental part of the database, and an incremental transaction dataset DN is constructed according to the number of transaction items in the transaction and the unit profit of the items, where DN = {T1, T2,..., T n , T n+1 ,..., T m}; There are m transactions in the incremental transaction data set DN, of which the n+1th to mth transactions are incremental parts;

[0104] S7), screening the incremental transaction data set, and screening out new candidate items from the candidate items in step S2); specifically:

[0105] S71), locate the incremental transaction part of the incremental transaction data set, and calculate the maxPer and TWU values ​​of each candidate item in the set I* of candidate items obtained in step S2);

[0106] S72), define the minimum utility threshold minutil and the maximum period threshold maxPer according to user needs, filter out new candidate items that meet these two thresholds, and sort the new candidate items in ascending order according to the size of the TWU value, and update the set I* of new candidate items.

[0107] S8) Scan the incremental transaction data set for the second time to expand the incremental utility list of new candidates; specifically:

[0108] S81), scanning the incremental transaction part of the incremental transaction data set again, and calculating the internal utility and remaining utility of the incremental transaction part corresponding to the candidate items in I*;

[0109] S82), expanding the incremental utility lists of these candidate items; expanding new tuples to the corresponding incremental utility lists stored in the LUL;

[0110] S83), deleting the incremental utility lists corresponding to the items not in I* from the global list (LUL) of the single incremental utility list, and sorting the LUL according to the order of I*;

[0111] S84) Update the remaining utility value of each incremental utility list in LUL. In this embodiment, the internal utility of the transactions in the original transaction data set D will not change with the arrival of incremental transactions. However, the remaining utility changes due to the change in the order of the items. Therefore, the following remaining utility maintenance strategy is given:

[0112] By reconstructing the incremental utility list of the new candidate single item, the saved information can be directly mined without scanning the incremental transaction data set again; the created global list is used to guide the connection order of subsequent incremental utility lists, as well as the reconstruction of related incremental utility lists after subsequent incremental transactions are added to the database.

[0113] S9), screening the incremental utility list of new candidate items based on a minimum utility threshold defined by a user to obtain periodic high utility items of the incremental database;

[0114] S10), using the pruning strategy in step S5) to screen new candidate items, and performing a connection operation on the incremental utility lists of the screened new candidate items to obtain a periodic high-utility item set of the incremental database;

[0115] S11), for new transactions that are continuously added, steps S6) to S10) are periodically executed to obtain the latest periodic high-utility item set of the incremental database.

[0116] Example 2

[0117] As shown in Table 1 and Table 2, it is assumed that the data set contains three elements: Tid, Transaction and TU. Tid represents the identifier. Transaction represents the transaction, which mainly includes the items and the number of items included in the transaction. Different transactions have different identifiers. TU represents the utility of the transaction. Uuit Profit represents the profit of the item.

[0118] According to the user's needs, the minimum threshold minutil=23 and maxPer=2 are set, and mining is performed in combination with the method of embodiment 1 of the present invention.

[0119] First, this embodiment scans the original database to obtain the original transaction data set D, such as the first five transactions in Table 1. Scan the data set D to obtain the TWU values ​​of the items as shown in the upper half of Table 3 and the maximum cycle value of each item. According to the TWU pruning strategy and the maximum cycle pruning strategy, the candidate items are sorted in ascending order of TWU as I*={f,a,d,e,g}.

[0120] Scan the dataset D for the second time and construct the incremental utility list iUL of all candidate items in I*, as shown in the upper part of Table 5. Then insert the candidate items into LUL according to their order, as shown in the upper part of Table 6, and then construct EUCS, as shown in the upper part of Table 4;

[0121] Execute the recursive process, starting from iUL(f). The sum of its iutil is 12, which is less than minutil, so it is not PHUI. However, the sum of its iutil and rutil is 39, which is not less than minutil, so consider its expanded item set. The subsequent items of item f are {a, d, e, g}, where the value of EUCS(f, a) is 21, which is less than minutil, so there is no need to construct the iUL of the item set {f, a}, nor to put a into the utility list set of the expanded item set of f. For {f, d}, {f, e}, and {f, g}, we construct their corresponding iUL and put them into the utility list set of the expanded item set of f. Then, the recursive program traverses depth to the item set {f, d}, and its corresponding

[0122] iUL({f,d})={(2,11,2),(3,10,6)},maxper=3. The pruning strategy that does not meet the sum of maxPer and iutil is screened, terminated and returned to the previous recursive layer.

[0123] Similarly, the subsequent {f,e} and {f,g} cannot pass the maximum cycle pruning strategy, and the corresponding recursive program ends, so the recursive program of item f ends. Next, the recursive program of item a is executed, and the final deep traversal obtains the item set {a,e} as PHUI. It will not be described in detail later, and the subsequent PHUIs are finally obtained as {d,g}. Therefore, the PHUI set of data set D is {{a,e},{d,g}};

[0124] As the incremental transaction set N is inserted, we scan part N to obtain the incremental transaction data set DN, as shown in the whole of Table 1. Scan the incremental part of the data set DN, accumulate the TWU value to the items in I*, and calculate the maxper value of the items in I*. Finally, prune item f and sort it in ascending order by TWU to obtain a new I* set of {e, g, a, d}. In addition, the order of LUL also changes accordingly. Then maintain LUL and recalculate the rutil value of the original transaction part. Scan the incremental part of the data set DN for the second time, and perform additional operations on the corresponding iUL in I*, as shown in the lower half of Table 5. Thus, the maintained LUL is obtained, as shown in the lower half of Table 6. Reconstruct EUCS, as shown in the lower half of the table. Finally, execute the recursive procedure, starting from item e. The recursive process is not described in detail here. Finally, the PHUI set of the data set DN is {{a}, {d}}.

[0125] Subsequently, the incremental transaction insertion is used as a trigger condition to repeat the process of step 4 of the example.

[0126] Table 1 Transaction dataset

[0127]

[0128] Table 2 Profit value of a single item

[0129] Item a b c d e f g Unit profit 5 3 1 4 2 3 2

[0130] Table 3 TWU values ​​of single items

[0131]

[0132] Table 4 EUCS table

[0133]

[0134] Table 5 Incremental utility list

[0135]

[0136] Table 6 Global List

[0137]

[0138] Example 3

[0139] like Figure 3 As shown, this embodiment provides a system for mining periodic utility itemsets in an incremental database, and the system includes:

[0140] A data acquisition module is used to scan the transaction information of the original database and the subsequent incremental database to obtain the original transaction data set or the incremental transaction data set;

[0141] The data screening module is used to screen the transaction data set to obtain candidate items;

[0142] The utility list construction and maintenance module is used to construct the utility list of the corresponding items for the candidate items, obtain the corresponding utility list, and maintain the incremental part;

[0143] The utility list merging module is used to merge the utility lists corresponding to single items or item sets that meet the pruning strategy, and determine whether they are periodic high-utility item sets.

[0144] Example 4

[0145] This embodiment provides a periodic utility item set mining device under an incremental database, comprising:

[0146] at least one processor;

[0147] at least one memory for storing at least one program;

[0148] When at least one of the programs is executed by at least one of the processors, the at least one processor implements the mining method.

[0149] The above embodiments and descriptions are only for illustrating the principles and best embodiments of the present invention. Without departing from the spirit and scope of the present invention, the present invention may be subject to various changes and improvements, all of which fall within the scope of the present invention to be protected.

Claims

1. A method for mining periodic utility itemsets in an incremental database, characterized in that: The following steps are involved: S1), scanning the transaction information of the original database to obtain the original transaction data set; S2) Use the pruning strategy to filter the original transaction data set to obtain candidate items, which specifically includes the following steps: S21), calculating the TWU value and the maximum period mPer value of each item in the original transaction data set; S22), defining a minimum utility threshold minutil and a maximum period threshold maxPer according to user needs, screening out candidate items that meet these two thresholds, and sorting the candidate items in ascending order according to the size of the TWU value, and obtaining a set of candidate items I*; S3) Scan the original transaction data set for the second time to obtain the incremental utility list of candidate items, which specifically includes the following steps: S31), scan the original transaction database again, and calculate the internal utility of all candidate items in the candidate item set I*; S32), calculate the remaining utility of all candidate items; S33), constructing an incremental utility list of candidate items according to the internal utility and remaining utility of the candidate items; S34), inserting the incremental utility lists of all candidate items into the global list LUL of the single incremental utility list, and sorting them in the order of I*; S4), in the global list LUL, the incremental utility list of candidate items is screened based on the minimum utility threshold defined by the user to obtain the periodic high utility items of the original database; S5), in the global list LUL, the candidate items are screened by using the pruning strategy, and the incremental utility lists of the screened candidate items are connected to obtain the periodic high-utility item set of the original database; S6), when a new transaction occurs, the database record increases, and the incremental part of the database is scanned to obtain the incremental transaction data set; S7), screening the incremental transaction data set, and screening out new candidate items from the candidate items in step S2), specifically including the following steps: S71), locate the incremental transaction part of the incremental transaction data set, and calculate the maximum period mPer and TWU value of each candidate item in the set I* of candidate items obtained in step S2); S72), defining a minimum utility threshold minutil and a maximum period threshold maxPer according to user needs, screening out new candidate items that meet the two thresholds, and sorting the new candidate items in ascending order according to the size of the TWU value, and updating to obtain a set I* of new candidate items; S8) Scan the incremental transaction data set for the second time to expand the incremental utility list of new candidate items, which specifically includes the following steps: S81), scanning the incremental transaction part of the incremental transaction data set again, and calculating the internal utility and remaining utility of the incremental transaction part corresponding to the candidate items in the updated I*; S82), expanding the incremental utility list of the candidate items, and correspondingly expanding the corresponding incremental utility list stored in the LUL with new tuples; S83) Delete the incremental utility lists corresponding to the candidates not in the updated I* from the global list LUL of the single-item incremental utility lists, and sort LUL according to the order of the updated I*. S84) Update the remaining utility values of each incremental utility list in the global list LUL. S9) In the global list LUL, screen the incremental utility lists of the new candidates based on the user-defined minimum utility threshold to obtain the periodic high-utility items of the incremental database. S10) In the global list LUL, use the pruning strategy in step S5) to screen the new candidates, and perform a join operation on the incremental utility lists of the screened new candidates to obtain the periodic high-utility item set of the incremental database. S11) For the continuously added new transactions, regularly execute steps S6) to S10) to obtain the latest periodic high-utility item set of the incremental database.

2. The method for mining periodic utility itemsets in an incremental database according to claim 1, characterized in that: In step S1), relevant information is extracted from the original database, and an original data set D is constructed according to the number of transaction items in the transaction and the unit profit of the items, where D = {T1, T2, ..., T n }, which means there are n transactions, each of which represents a transaction at a certain time, including the sale of several items; the utility calculation method of a single item is as follows: U(i,T c )=q(i,T c )*p(i); In the formula, U(i,T c ) indicates transaction T c The utility value of item set i, q(i,T c ) indicates transaction T c The number of item sets i in , p(i) represents the unit profit of item set i.

3. The method for mining periodic utility itemsets in an incremental database according to claim 1, characterized in that: In step S4), it specifically includes the following steps: S41) Calculate the utility value of the candidate according to the information of the incremental utility list. S42) Based on the user-defined minimum utility threshold minutil, compare the utility value of the candidate. If the utility value of this item is not less than the minimum utility threshold minutil, then this item is a periodic high-utility item, and the periodic high-utility item is obtained.

4. The method for mining periodic utility itemsets in an incremental database according to claim 1, characterized in that: In step S5), it specifically includes the following steps: S51) Calculate the sum of the total internal utility sumIutil and the total remaining utility sumRutil of the candidate according to the information of the incremental utility list. S52) Based on the user-defined minimum utility threshold minutil, compare the sum of the total internal utility sumIutil and the total remaining utility sumRutil. If sumIutil(X)+sumRutil(X)<minutil, then prune this item set X and its extended item sets. Otherwise, obtain the screened candidates. S53) For the incremental utility lists corresponding to these screened candidates, perform a depth-first traversal for expansion to obtain the periodic high-utility item set of the original database.

5. A system for the method for mining periodic utility itemsets in an incremental database according to any one of claims 1 to 4, characterized in that: The described system includes: A data acquisition module, used to scan the transaction information of the original database and the subsequent incremental database to obtain the original transaction data set or the incremental transaction data set. A data screening module, used to screen the transaction data set to obtain candidates. A utility list construction and maintenance module, used to construct the utility list of the corresponding items for the candidates to obtain the corresponding utility list, and to maintain the incremental part. A utility list merging module, used to merge the utility lists corresponding to the single items or item sets that meet the pruning strategy, and to judge whether it is a periodic high-utility item set.

6. A device for the method for mining periodic utility itemsets in an incremental database according to any one of claims 1 to 4, characterized in that: The described device includes: At least one processor; At least one memory, used to store at least one program; When at least one of the at least one program is executed by at least one of the at least one processor, so that at least one of the at least one processor implements the mining method described in any one of claims 1-4.

Images