A COMPARATIVE STUDY OF FREQUENT ITEMSET MINING ALGORITHMS

SufRec, an algorithm for mining association rules: Recursivity and task parallelism

Since the era of data explosion, data mining in transactional databases has become more and more important. There are many data mining techniques like association rule mining, the most important and well-researched one. Furthermore, frequent itemset mining is one of the fundamental but time-consuming steps in association rule mining. Most of the algorithms used in literature find frequent itemsets on search space items having at least a minsup and are not reused for mining next time. To deal with this problem, NOV-FI algorithms are proposed as a new approach in order to quickly detect frequent itemsets from transactional databases using an array of co-occurrences and occurrences of kernel item in at least one transaction. NOV-FI algorithms are easily expanded in distributed systems. Finally, the experimental results show that the proposed algorithms perform better than other existing algorithms.

The Association rule mining is one of the recent data mining research. Mining frequent itemsets in relational databases using relational queries give great attention to researchers nowadays. This paper implements modified set oriented algorithm for mining frequent itemsets in relational databases. In this paper, the sort and merge scan algorithm SETM (Houtsma and Swami, IEEE 25–33 (1995)) [1] is implemented for super market data set which is further improved by integrating transaction reduction technique. Our proposed algorithm Improved SETM (ISETM) generate the frequent itemsets from the database and find its execution time. Finally the performance of the algorithm is compared with the traditional Apriori and SETM algorithm.

This paper shows a mathematical foundation for almost important features in the problem of discovering knowledge by association rules. The class of frequent itemsets and the association rule set are partitioned into disjoint classes by two equivalence relations based on closures. Thanks to these partitions, efficient parallel algorithms for mining frequent itemsets and association rules can be obtained. Practically, one can mine frequent itemsets as well as association rules just in the classes that users take care of. Then, we obtain structures of each rule class using corresponding order relations. For a given relation, each rule class splits into two subsets of basic and consequence. The basic one contains minimal rules and the consequence one includes in the rules that can be deducted from those minimal rules. In the rest, we consider association rule mining based on order relation min. The explicit form of minimal rules according to that relation is shown. Due to unique representations of frequent itemsets through their generators and corresponding eliminable itemsets, operators for deducting all remaining rules are also suggested. Experimental results show that mining association rules based on relation min is better than the ones based on relations of minmin and minMax in terms of reduction in mining times as well as number of basic rules.

Frequent itemset mining and association rule mining are fundamental problems in data mining. Despite of the intensive and continuous researches on frequent itemset mining, one essential and not completely solved drawback still remains. The drawback is pattern explosion: a huge number of frequent itemsets, or patterns, will be often derived. The association rule mining takes over the same drawback since each association rule consists of two frequent itemsets. One promising solution for alleviating this essential drawback practically as a post-processing is to rank the patterns and rules from various aspects and to identify which one to be examined first. In this paper, we propose to utilize the technology of representation learning to rank frequent itemsets and association rules. More specifically, we develop several evaluation measures for frequent itemsets and association rules by using the vector representations of items, transactions and frequent patterns which are obtained by existing representation learning techniques. By employing vector representations of three kinds of objects, various rankings reflecting different aspects can expect to be obtained.

Frequent item set mining (FIM) algorithms extract subsets of items that occurs frequently in a collection of sets. FIM is a key analysis in several data mining applications, and the FIM tools are among the most computationally intensive data mining ones. In this work we present a many-core parallel version of a state-of-the-art FIM algorithm, DCI, whose sequential version resulted, for most of the tested datasets, better than FP-Growth, one of the most efficient algorithms for FIM. We propose a couple of parallelization strategies for Graphics Processing Units (GPU) suitable for different resource availability, and we present the results of several experiments conducted on real-world and synthetic datasets.

Frequent itemset mining is an important problem in the data mining area. Extensive efforts have been devoted to developing efficient algorithms for mining frequent itemsets. However, not much attention is paid on managing the large collection of frequent itemsets produced by these algorithms for subsequent analysis and for user exploration. In this paper, we study three structures for indexing and querying frequent itemsets: inverted files, signature files and CFP-tree. The first two structures have been widely used for indexing general set-valued data. We make some modifications to make them more suitable for indexing frequent itemsets. The CFP-tree structure is specially designed for storing frequent itemsets. We add a pruning technique based on length-2 frequent itemsets to make it more efficient for processing superset queries. We study the performance of the three structures in supporting five types of containment queries: exact match, subset/superset search and immediate subset/superset search. Our results show that no structure can outperform other structures for all the five types of queries on all the datasets. CFP-tree shows better overall performance than the other two structures.

Online mining of data streams is an important data mining problem with broad applications. However, it is also a difficult problem since the streaming data possess some inherent characteristics. In this paper, we propose a new single-pass algorithm, called DSM-FI (data stream mining for frequent itemsets), for online incremental mining of frequent itemsets over a continuous stream of online transactions. According to the proposed algorithm, each transaction of the stream is projected into a set of sub-transactions, and these sub-transactions are inserted into a new in-memory summary data structure, called SFI-forest (summary frequent itemset forest) for maintaining the set of all frequent itemsets embedded in the transaction data stream generated so far. Finally, the set of all frequent itemsets is determined from the current SFI-forest. Theoretical analysis and experimental studies show that the proposed DSM-FI algorithm uses stable memory, makes only one pass over an online transactional data stream, and outperforms the existing algorithms of one-pass mining of frequent itemsets.

Frequent pattern mining is the most important phase of association rule mining process because of its time and space complexity. Several methods have attempted to improve the performance of association rule mining by enhancing frequent pattern mining efficiency. Due to the large size of the data-sets and huge amounts of data which should be mined, many parallel and distributed mining approaches have been introduced to divide data-sets or to distribute mining processes between multiple processors or computers and thus, improve the efficiency of the mining process. In this paper, we propose a hadoop-based parallel implementation of PrePost+ algorithm for frequent itemset mining. In our parallel approach, the process of constructing N-Lists of itemsets has been distributed between the mappers and the operation of the final pruning process and extracting frequent itemsets has been carried out by reducers in a map-reduce parallel programming model. The experimental results show that our hadoop-based PrePost+(HBPrePost+) algorithm outperforms one of the best existing parallel methods of frequent itemset mining (PARMA) in terms of execution time.

Mining frequent itemsets in databases is an important and widely studied problem in data mining research. The problem of mining frequent itemsets is usually solved by constructing candidates of itemsets, and identifying those itemsets that meet the requirement of frequent itemsets. This paper proposes a novel algorithm based on BitTable (or bitmap) representation of the data. Data - related to frequent itemsets - are stored in spare matrices. Simple matrix and vector multiplications are used to calculate the support of the potential n+1 itemsets. The main benefit of this approach is that only bitmaps of the frequent itemsets are generated. The concept is simple and easily interpretable and it supports a compact and effective implementation (in MATLAB). An application example related to the BMS-WebView-1 benchmark data is presented to illustrate the applicability of the proposed algorithm.

Association rule mining is a particularly well studied field in data mining given its importance as a building block in many data analytics tasks. Many studies have focused on efficiency because the data to be mined is typically very large. However, while there are many approaches in literature, each approach claims to be the fastest for some given dataset. In other words, there is no clear winner. On the other hand, there is panoply of algorithms and implementations specifically designed for parallel computing. These solutions are typically implementations of sequential algorithms in a multi-processor configuration focusing on load balancing and data partitioning, each processor running the same implementation on it is own partition. The question we ask in this paper is whether there is a means to select the appropriate frequent itemset mining algorithm given a dataset and if each processor in a parallel implementation could select its own algorithm provided a given partition of the data.

The step of mining frequent itemsets in database is the essential step and most expensive in the process of mining association rules in data mining task, many algorithms of mining frequent itemsets have been proposed to improve the performance of Apriori Algorithm. In this paper, we have introduced an optimization in the phase of generation pruning of candidates by a new strategy for the calculation of frequent itemsets based on approximate values of supports exact the itemsets. We have evaluated our algorithm AprioriMin against three popular frequent itemsets mining algorithms — Apriori and FP-growth, Close using two data sets with a variety of minimum support.

Efficiently mining frequent itemsets and association rules on the encrypted outsourced data remains a great challenge for the time-consuming ciphertext computations. Nowadays, it has been not well addressed for privacy-preserving frequent itemsets and association rule mining schemes with mining efficiency, dataset, and query confidentiality simultaneously. In this paper, we investigate the study of privacy issues on frequent itemset mining and association rule mining on outsourced data in a two-cloud model, where the data are encrypted and outsourced by multiple owners holding different public keys. We develop several secure computation protocols based on additively homomorphic cryptosystem and additive secret sharing, which enable the clouds could securely mine the frequent itemsets and association rules. Furthermore, we also design two kinds of frequent itemset and association rule query service models, i.e., service customers query the cloud-mined results, and service customers query with their own decided threshold. The proposed scheme not only supports the mining process on the data encrypted by multiple public keys without compromising the security of the datasets, query data and query results, but also offline users. In addition, the experimental results show that our query scheme is much more efficient than the state-of-the-art work.

Association rules mining is an important and classic research topic in Data Mining, and has been widely applied in many real-life cases. The primary time and memory consumption in association rules mining is from its first step - frequent itemsets mining. With the development of cloud computing, outsourcing this task to third-party service providers will save efforts in system development, deployment, operation, etc. Outsourcing, however, actually brings risks and difficulties in verifying the results returned by these services. In this paper, we focus on verifying the integrity of the results returned by outsourcing services. We propose a metamorphic-based method, which is light-weight and requires not much complicated process. The key point of our method is the construction of a set of metamorphic relations (MRs). Through analysis and experimental research, we show that our approach delivers quite satisfactory results.

Finding the frequent itemset is one of the most investigated extents of data mining. The Apriori algorithm is the most established algorithm for frequent itemset mining, but it has issues regarding scanning frequent databases and generating a large amount of candidate sets. To solve these issues, an Improved Apriori algorithm was proposed. We examined the data structure, implementation, and algorithmic features that mainly focus on frequent itemset mining. We are representing an Improved Apriori algorithm on Spark in which simple and scalable implementation is done to achieve a faster process with lower support thresholds. We examined the improved Apriori algorithm on Extended Bakery Dataset and Retail Dataset. The results show execution time was reduced by 40% and 57% compared with the original Apriori algorithm.