Towards Efficient Re-mining of Frequent Patterns upon Threshold Changes

Closed frequent similar pattern mining: Reducing the number of frequent similar patterns without information loss

Efficient algorithms for incremental maintenance of closed sequential patterns in large databases

Abstract. Frequent sequential pattern mining in item interval extended sequence database (iSDB) has been one of interesting task in recent years. Unlike classic frequent sequential pattern mining, the pattern mining in iSDB also consider the item interval between successive items; thus, it may extract more meaningful sequential patterns in real life. Most previous frequent sequential pattern mining in iSDB algorithms needs a minimum support threshold (minsup) to perform the mining. However, it’s not easy for users to provide an appropriate threshold in practice. The too high minsup value will lead to missing valuable patterns, while the too low minsup value may generate too many useless patterns. To address this problem, we propose an algorithm: TopKWFP – Top-k weighted frequent sequential pattern mining in item interval extended sequence database. Our algorithm doesn’t need to provide a fixed minsup value, this minsup value will dynamically raise during the mining process

Techniques for mining rare patterns have been researched in the association rule mining area because traditional frequent pattern mining methods have to generate a large amount of unnecessary patterns in order to find rare patterns from large databases. One such technique, the multiple minimum support threshold framework was devised to extract rare patterns by using a different minimum item support threshold for each item in a database. Nevertheless, this framework cannot sufficiently reflect environments of the real world. The reason is that it does not consider weights of items, such as market prices of products and fatality rates of diseases, in its mining process. Therefore, an algorithm has been proposed to mine rare patterns with utilities exceeding a user-specified minimum utility by considering rarity and utility information of items. However, since this algorithm employs the concept of traditional high utility pattern mining, patterns' lengths are not considered for determining utilities of the patterns. If the length of a pattern is sufficiently long, the pattern is more likely to have an enough utility to become a high utility pattern regardless of item utilities within the pattern. Therefore, the algorithm cannot guarantee that all items in a mined pattern have high utilities. In this paper, we propose a novel algorithm that effectively reduces such dependency of patterns on their lengths by considering their lengths in the mining process in order to mine more meaningful rare patterns compared to patterns mined by previous algorithms. Experimental results demonstrate that our algorithm extracts a lesser number of more meaningful patterns and consumes less computational resources compared to state-of-the-art algorithms.

Frequent pattern mining is a classic data mining technique, generally applicable to a wide range of application domains, and a mature area of research. The fundamental challenge arises from the combinatorial nature of frequent itemsets, scaling exponentially with respect to the number of unique items. Apriori-based and FPTree-based algorithms have dominated the space thus far. Initial phases of this research relied on the Apriori algorithm and utilized a distributed computing environment; we proposed the Cartesian Scheduler to manage Apriori's candidate generation process. To address the limitation of bottom-up frequent pattern mining algorithms such as Apriori and FPGrowth, we propose the Frequent Hierarchical Pattern Tree (FHPTree): a tree structure and new frequent pattern mining paradigm. The classic problem is redefined as frequent hierarchical pattern mining where the goal is to detect frequent maximal pattern covers. Under the proposed paradigm, compressed representations of maximal patterns are mined using a top-down FHPTree traversal, FHPGrowth, which detects large patterns before their subsets, thus yielding significant reductions in computation time. The FHPTree memory footprint is small; the number of nodes in the structure scales linearly with respect to the number of unique items. Additionally, the FHPTree serves as a persistent, dynamic data structure to index frequent patterns and enable efficient searches. When the search space is exponential, efficient targeted mining capabilities are paramount; this is one of the key contributions of the FHPTree. This dissertation will demonstrate the performance of FHPGrowth, achieving a 300x speed up over state-of-the-art maximal pattern mining algorithms and approximately a 2400x speedup when utilizing FHPGrowth in a distributed computing environment. In addition, we allude to future research opportunities, and suggest various modifications to further optimize the FHPTree and FHPGrowth. Moreover, the methods we offer will have an impact on other data mining research areas including contrast set mining as well as spatial and temporal mining.

Mining patterns in large databases is a challenging task facing NP-hard problems. Research focused attention on the most occurrent patterns, although less frequent patterns still offer interesting insights. In this paper we propose a new algorithm for discovering infrequent patterns and compare it to other solutions.

Frequent graph mining allows us to find useful frequent sub-graph patterns from large and complicated graph databases. In a lot of real world applications, graph patterns with relatively low supports can be used as meaningful information. However, previous methods based on a single minimum support threshold have trouble finding them. That is, they cause “rare item problem”, which means that useful sub-graphs with low supports cannot be extracted when a minimum support threshold is high, while an enormous number of patterns have to be mined to obtain these useful ones when the value is low. To overcome this problem, we propose a novel algorithm, FGM-MMS (Frequent Graph Mining based on Multiple Minimum Support constraints). After that, we demonstrate that the suggested algorithm outperforms a state-of-the-art graph mining algorithm through comprehensive performance experiments.

Mining frequent inter-transaction patterns (ITPs) from large databases is both useful and of interest. Since frequent inter-transaction patterns (FITPs) are discovered across transactions in a transaction database (TD), the number of patterns is very large. Therefore, the mining time and memory usage are very high. Although several algorithms have been proposed for mining FITPs, they still require long runtime and high memory usage. Recent research shows that N-list-based approaches are very efficient for mining frequent patterns (FPs). Therefore, in this paper, we propose an N-list-based algorithm, called NL-ITP-Miner, to mine FITPs. In the proposed algorithm, we adopt the advantages of the N-list structure to build up the IT-PPC-tree. During the process of building the IT-PPC-tree, NL-ITP-Miner applies our proposed theorems to eliminate infrequent inter-transaction 1-patterns to reduce the search space. NL-ITP-Miner scans the database once to find frequent inter-transaction (FIT) 1-patterns for constructing the IT-PPC-tree, after that, the NL-ITP-Miner algorithm traverses this tree to generate frequent 1-patterns, FIT 1-patterns with their respective N-lists. Besides, we also propose effective pruning strategies that help NL-ITP-Miner to reduce the search space significantly and generate FITPs more quickly. Experiments show that NL-ITP-Miner outperforms the state-of-the-art algorithms for mining FITPs in terms of runtime and memory usage.

Objectives: To develop a memory efficient, incremental and interactive distributed FPM having less communication and synchronization overhead with good load balancing capability, to analyze the dynamic transactional data in a distributed database. Methods/Analysis: This technique adopts prefix based equivalence class partitioning scheme to generate frequent item sets without generating local frequent sets with low memory consumption. This approach uses a range of support values to update the frequent patterns with less time complexity. This paper proposes distributed FPM techniques with both count distributed and compressed data distributed parallel approaches. The performance of the algorithms are tested and compared with popular distributed FPM algorithms using standard datasets. Findings: To deal with the massive dynamic data stored in distributed databases, this approach develops three distributed frequent set generation algorithms, which update frequent patterns by reusing the previously stored pattern information with no complex calculations or data structures. The proposed approaches also provide the user with the facility to interactively adjust the minimum support value as per their own conveniences by keeping the nearly frequent itemsets with the help of two minimum support thresholds (low, high). Measures have been taken to reduce the additional itemset storage and computations as well as to achieve good load balancing with low communication and synchronization overhead. Since the proposed algorithms adopt prefix based equivalent class partitioning technique at each n-itemset level and undergo four levels of itemset filtering to remove infrequent items from each class before calculating the individual item count, the inter node communication required is less in this approach. To eliminate the drawbacks of both count and data distribution approaches one of the algorithms proposed adopts a hybrid approach which distributes the compressed data only once, hence communication overhead is less compared with other DD algorithms. Conclusion/Application: The proposed distributed techniques reduce memory utilization and itemset comparisons compared to the existing approaches. The performances are tested and evaluated for market analysis and online credit card fraud detection applications.

Frequent pattern mining is a basic problem in data mining and knowledge discovery. The discovered patterns can be used as the input for analyzing association rules, mining sequential patterns, recognizing clusters, and so on. However, discovering frequent patterns in large scale datasets is an extremely time consuming task. Most research in the area of association rule discovery has focused on the method of efficient frequent pattern discovery, e.g. Park, Chen & Yu (1995); Savasere, Omiecinski & Navathe (1995); Han, Pei & Yin (2000); Pei, Han, Lu, Nishio, Tang & Yang (2001). When seeking all associations that satisfy constraints on support and confidence, once frequent patterns have been identified, generating the association rules is trivial. In the last decade, various algorithms have been proposed on this problem, e.g. maximal pattern mining - Grahne & Zhu (2003); closed pattern mining - Pei, Han & Mao (2000); Grahne & Zhu (2003); mining the most interesting frequent patterns - Fu, Kwong & Tang (2000); Han, Wang, Lu & Tzvetkov (2002, 2005); Hirate, Iwahashi & Yamana (2004). However, some challenges are still existed and need to be overcome. In this paper, a mathematical space will be introduced with some new related concepts and propositions to design a new algorithm for solving frequent patterns mining problem. It is hoped that such an improved algorithm will be simple to implement and more efficient.

Recent research in frequent pattern mining (FPM) has shifted from obtaining the complete set of frequent patterns to generating only a representative (summary) subset of frequent patterns. Most of the existing approaches to this problem adopt a two-step solution; in the first step, they obtain all the frequent patterns, and in the second step, some form of clustering is used to obtain the summary pattern set. However, the twostep method is inefficient and sometimes infeasible since the first step itself may fail to finish in a reasonable amount of time. In this paper, we propose an alternative approach to mining frequent pattern representatives based on a uniform sampling of the output space. Our new algorithm, Musk, obtains representative patterns by sampling uniformly from the pool of all frequent maximal patterns; uniformity is achieved by a variant of Markov Chain Monte Carlo (MCMC) algorithm. Musk simulates a random walk on the frequent pattern partial order graph with a prescribed transition probability matrix, whose values are computed locally during the simulation. In the stationary distribution of the random walk, all maximal frequent pattern nodes in the partial order graph are sampled uniformly. Experiments on various kind of graph and itemset databases validate the effectiveness of our approach.

Existing studies for mining frequent XML query patterns mainly introduce a straightforward candidate generate-and-test strategy and compute frequencies of candidate query patterns from scratch periodically by checking the entire transaction database, which consists of XML query patterns transformed from user queries. However, it is nontrivial to maintain such discovered frequent patterns in real XML databases because there may incur frequent updates that may not only invalidate some existing frequent query patterns but also generate some new frequent ones. Accordingly, existing proposals are inefficient for the evolution of the transaction database. To address these problems, this paper presents an efficient algorithm IPS-FXQPMiner for mining frequent XML query patterns without candidate maintenance and costly tree-containment checking. We transform XML queries into sequences through a one- to-one mapping and then mine the frequent sequences to generate frequent XML query patterns. More importantly, based on IPS-FXQPMiner, an efficient incremental algorithm, Incre-FXQPMiner is proposed to incrementally mine frequent XML query patterns, which can minimize the I/O and computation requirements for handling incremental updates. Our experimental study on various real-life datasets demonstrates the efficiency and scalability of our algorithms over previous known alternatives.

Vertical itemset mining is an important frequent pattern mining problem with broad applications. It is challenging since one may need to examine a combinatorial explosive number of possible patterns of items of a dataset in a traditional horizontal algorithm. Since high dimensional datasets typically contain a large number of columns and a small number of rows, vertical itemset mining algorithms, which extract the frequent itemsets of dataset by producing all combination of rows ids, are a good alternative for horizontal algorithms in mining frequent itemsets from high dimensional dataset. Since a rowset can be simply produced from its subsets by adding a new row id to a sub rowset, many bottom up vertical itemset mining algorithms are designed and represented in the literature. However, bottom up vertical mining algorithms suffer from a main drawback. Bottom-up algorithms start the process of generating and testing of rowsets from the small rowsets and go on to the larger rowsets, whereas the small rowsets cannot produce a frequent itemsets because they contain less than minimum support threshold number of rows. In this paper, we described a new efficient vertical top down algorithm called VTD (Vertical Top Down) to conduct mining of frequent itemsets in high dimensional datasets. Our top down approach employed the minimum support threshold to prune the rowsets which any itemset could not be extracted from them. Several experiments on real bioinformatics datasets showed that VTD is orders of magnitude better than previous closed pattern mining algorithms. Our performance study showed that this algorithm outperformed substantially the best former algorithms.

: Discovery of sequential patterns is becoming increasingly useful and essential in many scientific and commercial domains. Enormous sizes of available datasets and possibly large number of mined patterns demand efficient, scalable, and parallel algorithms. Even though a number of algorithms have been developed to efficiently parallelize frequent pattern discovery algorithms that are based on the candidate-generation-and-counting framework, the problem of parallelizing the more efficient projection-based algorithms has received relatively little attention and existing parallel formulations have been targeted only toward shared-memory architectures. The irregular and unstructured nature of the task-graph generated by these algorithms and the fact that these tasks operate on overlapping sub-databases makes it challenging to efficiently parallelize these algorithms on scalable distributed-memory parallel computing architectures. In this paper we present and study a variety of distributed-memory parallel algorithms for a tree-projection-based frequent sequence discovery algorithm that are able to minimize the various overheads associated with load imbalance, database overlap, and interprocessor communication. Our experimental evaluation on a 32 processor IBM SP show that these algorithms are capable of achieving good speedups, substantially reducing the amount of the required work to find sequential patterns in large databases.

Clinical research has led to the development of new medications and medication strategy may change to gain better treatment outcomes. With the increasing attention to the evidence based medical guideline, it is important for clinicians to study the medication strategy changes. This study is interested in evaluating long-term prescriptions of type 2 diabetes patients provided by Kyoto University Hospital following the release of a new medication in 2010. Frequent sequential pattern mining (FSPM) is a prominent tools for extracting frequent patterns. However, the number of the result set could be enormous and may inhibit clinicians in assessing the results. To help the clinicians, a medication progression graph (MPG) is constructed using adjacent (1-sequence) frequent patterns produced by the singleton pattern mining. Compared to conventional frequent patterns, singleton's patterns features full itemsets and 1-step distance. Hence, the pattern represents the true medication transition event. Using the graph, a clinical physician was able to observe a significant increase of physician's activities in changing the medication after the release (i.e., in MPG that start with Sulfonylurea after 2010, the number of edges is increase by 2.83 and the number of nodes is increase by 2.27 compared to before 2010). Our preliminary results show that the new visualization method enables a clinician to analyze a medication strategy changes after the new medication released.