Frequent Itemset Mining Methods Research Articles

GMiner++: Boosting GPU-based frequent itemset mining by reducing redundant computations

Frequent itemset mining (FIM) is increasingly important in fundamental data mining techniques. However, the applicability of existing FIM methods is limited, mainly because of their performance. The expected performance improvement is limited owing to the exploitation of only a single thread, despite numerous efficient single-threaded FIM methods being proposed. Numerous parallel FIM methods have been devised using graphic processing units (GPU) or multicore central processing units (CPUs) to overcome the shortcomings of these methods. However, when extracting patterns from large amounts of data, multi-threaded FIM methods exhibit a similar performance tendency to single-threaded FIM methods, because of their large memory footprints and computations. Hence, we propose GMiner++, a memory-efficient GPU-based FIM method equipped with several GPUs. We propose a sub-database of the same size called bit array blocks, which contains pre-calculated bit arrays of F1∪P(IK). These bit arrays are repeatedly exploited during mining tasks using an elegant probabilistic model. GMiner++ can obtain frequent patterns and use several GPUs only by using the bit array blocks and the occurrence update scheme. The proposed method decreased redundant computations using pre-calculated bit arrays with bit array blocks. In addition, GMiner++ does not create intermediate data during mining tasks to increase robustness and reduce memory footprints. Simulation results demonstrate that GMiner++ outperformed existing FIM methods concerning performance and scalability with increasing robustness.

Detecting shilling groups in online recommender systems based on graph convolutional network

Online recommender systems have been shown to be vulnerable to group shilling attacks in which attackers of a shilling group collaboratively inject fake profiles with the aim of increasing or decreasing the frequency that particular items are recommended. Existing detection methods mainly use the frequent itemset (dense subgraph) mining or clustering method to generate candidate groups and then utilize the hand-crafted features to identify shilling groups. However, such two-stage detection methods have two limitations. On the one hand, due to the sensitivity of support threshold or clustering parameters setting, it is difficult to guarantee the quality of candidate groups generated. On the other hand, they all rely on manual feature engineering to extract detection features, which is costly and time-consuming. To address these two limitations, we present a shilling group detection method based on graph convolutional network. First, we model the given dataset as a graph by treating users as nodes and co-rating relations between users as edges. By assigning edge weights and filtering normal user relations, we obtain the suspicious user relation graph. Second, we use principal component analysis to refine the rating features of users and obtain the user feature matrix. Third, we design a three-layer graph convolutional network model with a neighbor filtering mechanism and perform user classification by combining both structure and rating features of users. Finally, we detect shilling groups through identifying target items rated by the attackers according to the user classification results. Extensive experiments show that the classification accuracy and detection performance (F1-measure) of the proposed method can reach 98.92% and 99.92% on the Netflix dataset and 93.18% and 92.41% on the Amazon dataset.

On a parallel spark workflow for frequent itemset mining based on array prefix‐tree

AbstractExtracting frequent itemsets from datasets is an important problem in data mining, for which several mining methods including FP‐Growth have been proposed. FP‐Growth is a classical frequent itemset mining method, which generates pattern databases without candidates. Many improvements have been made in the literature due to the high time complexity and memory usage of FP‐Growth. However, most of them still suffer from performance issues on large datasets. In this paper, we design an auxiliary structure, Array Prefix‐Tree (AP‐Tree), and propose a new algorithm, Array Prefix‐Tree Growth (APT‐Growth), which is further parallelized as a Spark workflow, referred to as PAPT‐Growth. Based on a density threshold, we incorporate an adaptive algorithm selection process into PAPT‐Growth to ensure its running time performance. We conduct extensive experiments on different thresholds and multiple datasets, and experimental results show the performance superiority of PAPT‐Growth in comparison with several state‐of‐the‐art methods such as PFP, YAFIM, and DFPS. The analysis on density reveals a changing point, which justifies the necessity and validity of adaptive algorithm selection.

Parallelization of Frequent Itemset Mining Methods with FP-tree: An Experiment with PrePost+ Algorithm

Parallel processing has turn to be a common programming practice because of its efficiency and thus becomes an interesting field for researchers. With the introduction of multi- core processors as well as general purpose graphics processing units, parallel programming has become affordable. This leads to the parallelization of many of the complex data processing algorithms including algorithms in data mining. In this paper, a study on parallel PrePost+ is presented. PrePost+ is an efficient frequent itemset mining algorithm. The algorithm has been modified as a parallel algorithm and the obtained result is compared with the result of sequential PrePost+ algorithm

Fast Eclat Algorithms Based on Minwise Hashing for Large Scale Transactions

The Eclat algorithm is one of the most widely used frequent itemset mining methods. In the normal Eclat algorithm and its variants, it is inefficient to calculate the intersection size of itemsets by sequentially comparing elements, especially for large scale transactions. In this paper, we propose the fast Eclat algorithms that can quickly calculate the intersection size of multiple itemsets by using minwise hashing and the estimators. Minwise hashing is used to calculate the Jaccard similarity coefficient by mapping the elements of the sets to those of smaller sets. Two estimators are used to estimate the intersection size of itemsets based on the Jaccard similarity coefficient. Due to the “imperfect” hash function, minwise hashing may obtain a biased Jaccard similarity, which results in error between the real value and the estimated value of the intersection size. Thus, we proposed the HashEclat which uses the maximum of ${|}{A}{|}$ and ${|}{B}{|}$ to represent the union size ${|}{A}{\cup }{B}{|}$ , and proposed the Sim-Eclat which uses the minimum of ${|A|}$ and ${|}{B}{|}$ to represent the intersection size ${|}{A}{\cap }{B}{|}$ . Furthermore, we use a boundary error ${E}$ for better performance as follows: if ${E}$ is large, the intersection size is determined by a traditional method, and the result is more accurate but takes longer to compute; otherwise, it will be the opposite. Both the theoretical analysis and experimental results show that the proposed algorithms can obtain almost all frequent itemsets with higher speed and less memory usage than other algorithms.

HashEclat: an efficient frequent itemset algorithm

The Eclat algorithm is one of the most widely used frequent itemset mining methods. However, the inefficiency for calculating the intersection of itemsets makes it a time-consuming method, especially when the dataset has a large number of transactions. In this work, for the purpose of efficiency improvement, we proposed an approximate Eclat algorithm named HashEclat based on MinHash, which could quickly estimate the size of the intersection set, and adjust the parameters k, E and minSup to consider the tradeoff between accuracy of the mining results and execution time. The parameter k is the top-k parameter of one-permutation MinHash algorithm; the parameter E is the estimate error of one intersection size; the parameter minSup is the minimum support threshold. In many real situations, an approximate result with faster speed maybe more useful than ‘exact’ result. The theoretical analysis and experiment results that we present demonstrate that the proposed algorithm can output almost all of the frequent itemset with faster speed and less memory space.

Finding tendencies in streaming data using Big Data frequent itemset mining

The amount of information generated in social media channels or economical/business transactions exceeds the usual bounds of static databases and is in continuous growing. In this work, we propose a frequent itemset mining method using sliding windows capable of extracting tendencies from continuous data flows. For that aim, we develop this method using Big Data technologies, in particular, using the Spark Streaming framework enabling distributing the computation along several clusters and thus improving the algorithm speed. The experimentation carried out shows the capability of our proposal and its scalability when massive amounts of data coming from streams are taken into account.

BIGMiner: a fast and scalable distributed frequent pattern miner for big data

Frequent itemset mining is widely used as a fundamental data mining technique. Recently, there have been proposed a number of MapReduce-based frequent itemset mining methods in order to overcome the limits on data size and speed of mining that sequential mining methods have. However, the existing MapReduce-based methods still do not have a good scalability due to high workload skewness, large intermediate data, and large network communication overhead. In this paper, we propose BIGMiner, a fast and scalable MapReduce-based frequent itemset mining method. BIGMiner generates equal-sized sub-databases called transaction chunks and performs support counting only based on transaction chunks and bitwise operations without generating and shuffling intermediate data. As a result, BIGMiner achieves very high scalability due to no workload skewness, no intermediate data, and small network communication overhead. Through extensive experiments using large-scale datasets of up to 6.5 billion transactions, we have shown that BIGMiner consistently and significantly outperforms the state-of-the-art methods without any memory problems.

GMiner: A fast GPU-based frequent itemset mining method for large-scale data

Frequent itemset mining is widely used as a fundamental data mining technique. However, as the data size increases, the relatively slow performances of the existing methods hinder its applicability. Although many sequential frequent itemset mining methods have been proposed, there is a clear limit to the performance that can be achieved using a single thread. To overcome this limitation, various parallel methods using multi-core CPU, multiple machine, or many-core graphic processing unit (GPU) approaches have been proposed. However, these methods still have drawbacks, including relatively slow performance, data size limitations, and poor scalability due to workload skewness. In this paper, we propose a fast GPU-based frequent itemset mining method called GMiner for large-scale data. GMiner achieves very fast performance by fully exploiting the computational power of GPUs and is suitable for large-scale data. The method performs mining tasks in a counterintuitive way: it mines the patterns from the first level of the enumeration tree rather than storing and utilizing the patterns at the intermediate levels of the tree. This approach is quite effective in terms of both performance and memory use in the GPU architecture. In addition, GMiner solves the workload skewness problem from which the existing parallel methods suffer; as a result, its performance increases almost linearly as the number of GPUs increases. Through extensive experiments, we demonstrate that GMiner significantly outperforms other representative sequential and parallel methods in most cases, by orders of magnitude on the tested benchmarks.

Tree-based frequent itemsets mining for analysis of life-satisfaction and loneliness of retired athletes

Life-satisfaction and loneliness are two key indicators of individual mental state, and their detailed analysis could help improve the resettlement policy of retired athletes. This paper proposes a tree-based frequent itemsets mining method to estimate the influence factors of the life-satisfaction and the loneliness of retired athletes. The basic situations of the retired athletes are collected by the questionnaires and transformed into the binary attributes. Then, an extend prefix tree is built for mining the frequent itemsets. The lift measure is employed to generate the association rules based on the obtained frequent itemsets and realize the rules prune. The actual survey data of 750 Chinese retired athletes are adopted for comparing the proposed method and the Apriori algorithm. Experimental results verify the effectiveness of the proposed method is higher. Moreover, the obtained rules show that the health condition, the education, the social insurance participation affect both the life-satisfaction and the loneliness of retired athletes, and the income only affect the life-satisfaction of retired athletes.

An Improved Apriori Algorithm Based on an Evolution-Communication Tissue-Like P System with Promoters and Inhibitors

Apriori algorithm, as a typical frequent itemsets mining method, can help researchers and practitioners discover implicit associations from large amounts of data. In this work, a fast Apriori algorithm, called ECTPPI-Apriori, for processing large datasets, is proposed, which is based on an evolution-communication tissue-like P system with promoters and inhibitors. The structure of the ECTPPI-Apriori algorithm is tissue-like and the evolution rules of the algorithm are object rewriting rules. The time complexity of ECTPPI-Apriori is substantially improved from that of the conventional Apriori algorithms. The results give some hints to improve conventional algorithms by using membrane computing models.

Multiobjective Adaptive Fuzzy C-Means Algorithm for Mining Frequent Closed Itemset: A Lattice Based Approach for Time Series Data

Of late, frequent itemset mining from the data bases has emerged as a highly challenging task. And, there is a feast of various approaches launched with an eye on extracting the frequent rules from the database. In the erstwhile frequent itemset mining technique, the frequent itemset are extracted from the time series data by means of Adaptive Fuzzy C-Means (AFCM) clustering, though it is saddled with several setbacks in the clustering procedure. This arises on account of the fact that in clustering, the number of itemset is mammoth, and hence the method consumes a lot of time in addition to spending enhanced scanning time. Therefore, with a view tackling these adverse features, an innovative frequent itemset mining method is suggested in this document. At the outset, the time sequence database values are grouped by means of the Multiobjective AFCM method. Subsequently, the frequent itemset are extracted from the grouped outcomes by employing the lattice based method. The novel technique is performed in the MATLAB platform and its excellence is appraised by means of rainfall database.

Identification and Prioritization of Relationships between Environmental Stressors and Adverse Human Health Impacts.

BackgroundThere are > 80,000 chemicals in commerce with few data available describing their impacts on human health. Biomonitoring surveys, such as the NHANES (National Health and Nutrition Examination Survey), offer one route to identifying possible relationships between environmental chemicals and health impacts, but sparse data and the complexity of traditional models make it difficult to leverage effectively.ObjectiveWe describe a workflow to efficiently and comprehensively evaluate and prioritize chemical–health impact relationships from the NHANES biomonitoring survey studies.MethodsUsing a frequent itemset mining (FIM) approach, we identified relationships between chemicals and health biomarkers and diseases.ResultsThe FIM method identified 7,848 relationships between 219 chemicals and 93 health outcomes/biomarkers. Two case studies used to evaluate the FIM rankings demonstrate that the FIM approach is able to identify published relationships. Because the relationships are derived from the vast majority of the chemicals monitored by NHANES, the resulting list of associations is appropriate for evaluating results from targeted data mining or identifying novel candidate relationships for more detailed investigation.ConclusionsBecause of the computational efficiency of the FIM method, all chemicals and health effects can be considered in a single analysis. The resulting list provides a comprehensive summary of the chemical/health co-occurrences from NHANES that are higher than expected by chance. This information enables ranking and prioritization on chemicals or health effects of interest for evaluation of published results and design of future studies.CitationBell SM, Edwards SW. 2015. Identification and prioritization of relationships between environmental stressors and adverse human health impacts. Environ Health Perspect 123:1193–1199; http://dx.doi.org/10.1289/ehp.1409138

Iterative sampling based frequent itemset mining for big data

Frequent pattern mining attracts extensive research interests over the past two decades: including mining frequent item sets from transactions, extracting frequent sequences from bio-arrays and detecting common subgraph from molecular structures. In the era of big data, the explosive data volume brings new challenges to frequent pattern mining: (1) Space complexity: both input data, intermediate results and the outputted patterns could be too large to fit into memory which prevents many algorithms from executing; (2) Time complexity: many existing approaches rely on exhaustive search or complicated data structures to mine frequent patterns which prove to be inapplicable for big data. To deal with these two challenges. we propose ISbFIM, an Iterative Sampling based Frequent Itemset Mining method. Rather than process the entire data set at once, ISbFIM samples computationally-manageable subsets and extracts frequent itemsets from these subsets. By repeating this process for a sufficient number of times, we can guarantee both theoretically and empirically that the frequent itemsets can be enumerated without running into a combinatorial explosion. ISbFIM can be easily parallelized and applied to mine item sets, sequences or structures. We implement a Map-Reduce version of ISbFIM to demonstrate its scalability on big data.

Anomalous Event Detection in Traffic Video Based on Sequential Temporal Patterns of Spatial Interval Events

Detection of anomalous events from video streams is a challenging problem in many video surveillance applications. One such application that has received significant attention from the computer vision community is traffic video surveillance. In this paper, a Lossy Count based Sequential Temporal Pattern mining approach (LC-STP) is proposed for detecting spatio-temporal abnormal events (such as a traffic violation at junction) from sequences of video streams. The proposed approach relies mainly on spatial abstractions of each object, mining frequent temporal patterns in a sequence of video frames to form a regular temporal pattern. In order to detect each object in every frame, the input video is first pre-processed by applying Gaussian Mixture Models. After the detection of foreground objects, the tracking is carried out using block motion estimation by the three-step search method. The primitive events of the object are represented by assigning spatial and temporal symbols corresponding to their location and time information. These primitive events are analyzed to form a temporal pattern in a sequence of video frames, representing temporal relation between various object’s primitive events. This is repeated for each window of sequences, and the support for temporal sequence is obtained based on LC-STP to discover regular patterns of normal events. Events deviating from these patterns are identified as anomalies. Unlike the traditional frequent item set mining methods, the proposed method generates maximal frequent patterns without candidate generation. Furthermore, experimental results show that the proposed method performs well and can detect video anomalies in real traffic video data.

An Approach to the Mining of User Focused Frequent Itemsets Based on Attention

Usually, non trivial network visiting behaviors implied in network visiting log can be treated as the frequent itemsets or association rules if data in networking log file are transformed into transaction and technologies on association rule can be used to mine those frequent itemsets which are focused by user or some application. To mine non trivial behaviors of network visiting effectively, an attention based frequent itemsets mining method is proposed in this paper. In our proposed method, properties of users focusing is described as attention set and the early selection model of attention as information filter is referenced in the design of our method. Experimental results show that our proposed method is faster than apriori algorithm on the mining of frequent itemsets which is focused by our attention.

Efficient frequent itemset mining methods over time-sensitive streams

Stream data arrives dynamically and rapidly, and the characteristics cannot be reflected by the traditional transaction-based sliding window; thus, the mining results are inaccurate. This paper focuses on this problem and constructs a timestamp-based sliding window model, which can be further converted into a transaction-based sliding window. Based on this model, an extended enumeration tree is developed to incrementally maintain the essential information. In our proposed frequent itemset mining algorithm, we introduce the type transforming bound to dynamically classify the itemsets into categories; thus, certain itemset processing can be deferred or ignored, that is, an itemset will not be handled unless its type transforming bounds reach a threshold; as a result, the computational pruning can be conducted. Nevertheless, it only guarantees the conditions to obtain accurate results, and thus cannot achieve the best performance. This problem is further improved in our approximate mining algorithm, in which we propose a heuristic rule-based strategy. Additionally, it can save more computational cost with a tolerable mining error. Theoretical analysis and experimental studies demonstrate that our proposed algorithms have high accuracy, spend less computational time and memory, and significantly outperform the baseline method and state-of-the-art algorithms.

A Novel Frequent Item-set Mining Method based on Frequent-pattern List

A Novel Frequent Item-set Mining Method based on Frequent-pattern List

DRFP-tree: disk-resident frequent pattern tree

Frequent itemset mining methods basically address time scalability and greatly rely on available physical memory. However, the size of real-world databases to be mined is exponentially increasing, and hence main memory size is a serious bottleneck of the existing methods. So, it is necessary to develop new methods that do not fully rely on physical memory; new methods that utilize the secondary storage in the mining process should be the target. This motivates the work described in this paper; we mainly propose (Disk Resident Frequent Pattern) DRFP-Growth as a disk based approach similar to FP-Growth. DRFP-growth uses DRFP-tree, which is treated exactly as FP-tree when constructed in main memory and gets into a modified structure when it turns into disk resident to overcome the main memory bottleneck. This way, we are able to mine for frequent itemsets from databases of arbitrary sizes without being restricted by the available physical memory. In other words, we initially try to mine the database using the original FP-growth; we expand into the secondary memory only if we run out of physical memory. So, DRFP-growth is very comparable to FP-growth for small databases and high support threshold values. On the other hand, using DRFP-growth, we are still able to mine huge databases for low support threshold values (the only limitation is the available secondary storage rather than physical memory). The reported test results demonstrate how the proposed approach succeeds for cases where main memory based approaches fail.

Multidimensional fuzzy partitioning of attribute ranges for mining quantitative data

The article suggests a partitioning algorithm for quantitative attributes to support the discovery of frequent fuzzy patterns among transactions containing such attributes. More precisely, we present a heuristic, multivariate, top-down partitioning algorithm that divides attribute ranges into such intervals that the discovered frequent sets are also dense, and thus probably more interesting to the user. Our approach is fuzzy, so that the derived intervals have fuzzy bounds, and thereby also the derived frequent sets are fuzzy. The crisp (nonfuzzy) case is obtained as a special case. We evaluate the goodness of the partitioning method by measuring the average and absolute information amounts of the obtained fuzzy frequent sets. For the mining task, any fuzzy frequent item set mining method can be used. Experiments show that the algorithm is able to do multidimensional partitioning in a balanced way, and the “interestingness” of the obtained frequent sets is quite high, especially for correlated attributes. © 2004 Wiley Periodicals, Inc. Int J Int Syst 19: 1111–1126, 2004.