Sequential Pattern Mining Algorithm Research Articles

A new framework for detecting weighted sequential patterns in large sequence databases

Sequential pattern mining is an essential research topic with broad applications which discovers the set of frequent subsequences satisfying a support threshold in a sequence database. The major problems of mining sequential patterns are that a huge set of sequential patterns are generated and the computation time is so high. Although efficient algorithms have been developed to tackle these problems, the performance of the algorithms dramatically degrades in case of mining long sequential patterns in dense databases or using low minimum supports. In addition, the algorithms may reduce the number of patterns but unimportant patterns are still found in the result patterns. It would be better if the unimportant patterns could be pruned first, resulting in fewer but important patterns after mining. In this paper, we suggest a new framework for mining weighted frequent patterns in which weight constraints are deeply pushed in sequential pattern mining. Previous sequential mining algorithms treat sequential patterns uniformly while real sequential patterns have different importance. In our approach, the weights of items are given according to the priority or importance. During the mining process, we consider not only supports but also weights of patterns. Based on the framework, we present a weighted sequential pattern mining algorithm (WSpan). To our knowledge, this is the first work to mine weighted sequential patterns. The experimental results show that WSpan detects fewer but important weighted sequential patterns in large sequence databases even with a low minimum threshold.

Smart support functions for sequential pattern mining

In real applications, transaction data typically contain quantitative attributes. Existing approaches and algorithms proposed for sequential pattern mining such as AprioriAll often assume Boolean attributes (i.e., quantitative values are simply interpreted/transformed as Boolean values). This article addresses the impact of varied quantitative attributes in sequential pattern mining. More specifically, we define alternate smart support functions for computing the support measure of candidate sequential patterns. A noticeable advantage of this work is that the proposed smart support functions can be smoothly integrated into the framework of existing sequential pattern mining algorithms. In the discussion of this article, we assume adoption of the well-known AprioriAll algorithm and discuss the incorporation of the proposed smart support functions into this framework. The expected mining results are believed better reflecting the particular interests of different user groups and thus are more satisfactory to the intended users.

Generalized Sequential Pattern Mining with Item Intervals

Sequential pattern mining is an important data mining method with broad applications that can extract frequent sequences while maintaining their order. However, it is important to identify item intervals of sequential patterns extracted by sequential pattern mining. For example, a sequence with a 1-day interval and a sequence with a 1-year interval are completely different; the former sequence may have some association, while the latter may not. To adopt item intervals, two approaches have been proposed for integration of item intervals with sequential pattern mining; (1) constraint-based mining and (2) extended sequence-based mining. However, although constraint-based mining approach avoids the extraction of sequences with non-interest time intervals such as too long intervals it has setbacks in that it is difficult to specify optimal constraints related to item interval, and users must re-execute constraint-based algorithms with changing constraint values. On the other hand, extended sequence-based mining approach does not need to specify constraints and re-execute. Since extended sequence-based mining approach cannot adopt any constraints based on time intervals, it may extract meaningless patterns, such as sequences with too long item intervals. This means these two approaches have not only advantages but also disadvantages. To solve this problem, in this paper, we generalize sequential pattern mining with item interval. The generalization includes three points; (a) a capability to handle two kinds of item interval measurement, item gap and time interval, (b) a capability to handle extended sequences which are defined by inserting pseudo items based on the interval itemization function, and (c) adopting four item interval constraints. Generalized sequential pattern mining is able to substitute all types of conventional sequential pattern mining algorithms with item intervals. Using Japanese earthquake data, we have confirmed that our proposed algorithm is able to extract sequential patterns with item interval, defined in a flexible manner by the interval itemization function.

Mining Sequential Patterns More Efficiently by Reducing the Cost of Scanning Sequence Databases

Sequential pattern mining is a useful technique used to discover frequent subsequences as patterns in a sequence database. Depending on the application, sequence databases vary by number of sequences, number of individual items, average length of sequences, and average length of potential patterns. In addition, to discover the necessary patterns in a sequence database, the support threshold may be set to different values. Thus, for a sequential pattern-mining algorithm, responsiveness should be achieved for all of these factors. For that purpose, we propose a candidate-driven pattern-growth sequential pattern-mining algorithm called FSPM (Fast Sequential Pattern Mining). A useful property of FSPM is that the sequential patterns concerning a user-specified item can be mined directly. Extensive experimental results show that, in most cases FSPM outperforms existing algorithms. An analytical performance study shows that it is the inherent potentiality of FSPM that makes it more effective.

A fast interactive sequential pattern mining algorithm

In order to reduce the computational and spatial complexity in rerunning algorithm of sequential patterns query, this paper proposes sequential patterns based and projection database based algorithm for fast interactive sequential patterns mining algorithm (FISP), in which the number of frequent items of the projection databases constructed by the correct mining which based on the previously mined sequences has been reduced. Furthermore, the algorithm's iterative running times are reduced greatly by, using global-threshold. The results of experiments testify that FISP outperforms PrefixSpan in interactive mining.

A Sequential Pattern Mining Algorithm Based on Large-Itemset Reuse

A Sequential Pattern Mining Algorithm Based on Large-Itemset Reuse

On the use of hierarchical information in sequential mining-based XML document similarity computation

Measuring the structural similarity among XML documents is the task of finding their semantic correspondence and is fundamental to many web-based applications. While there exist several methods to address the problem, the data mining approach seems to be a novel, interesting and promising one. It explores the idea of extracting paths from XML documents, encoding them as sequences and finding the maximal frequent sequences using the sequential pattern mining algorithms. In view of the deficiencies encountered by ignoring the hierarchical information in encoding the paths for mining, a new sequential pattern mining scheme for XML document similarity computation is proposed in this paper. It makes use of a preorder tree representation (PTR) to encode the XML tree’s paths so that both the semantics of the elements and the hierarchical structure of the document can be taken into account when computing the structural similarity among documents. In addition, it proposes a postprocessing step to reuse the mined patterns to estimate the similarity of unmatched elements so that another metric to qualify the similarity between XML documents can be introduced. Encouraging experimental results were obtained and reported.

TSP: Mining top-k closed sequential patterns

Sequential pattern mining has been studied extensively in the data mining community. Most previous studies require the specification of a min_support threshold for mining a complete set of sequential patterns satisfying the threshold. However, in practice, it is difficult for users to provide an appropriate min_support threshold. To overcome this difficulty, we propose an alternative mining task: mining top-k frequent closed sequential patterns of length no less than min_l, where k is the desired number of closed sequential patterns to be mined and min_l is the minimal length of each pattern. We mine the set of closed patterns because it is a compact representation of the complete set of frequent patterns. An efficient algorithm, called TSP, is developed for mining such patterns without min_support. Starting at (absolute) min_support=1, the algorithm makes use of the length constraint and the properties of top-k closed sequential patterns to perform dynamic support raising and projected database pruning. Our extensive performance study shows that TSP has high performance. In most cases, it outperforms the efficient closed sequential pattern-mining algorithm, CloSpan, even when the latter is running with the best tuned min_support threshold. Thus, we conclude that, for sequential pattern mining, mining top-k frequent closed sequential patterns without min_support is more preferable than the traditional min_support-based mining.

Mining sequential patterns by pattern-growth: the PrefixSpan approach

Sequential pattern mining is an important data mining problem with broad applications. However, it is also a difficult problem since the mining may have to generate or examine a combinatorially explosive number of intermediate subsequences. Most of the previously developed sequential pattern mining methods, such as GSP, explore a candidate generation-and-test approach [R. Agrawal et al. (1994)] to reduce the number of candidates to be examined. However, this approach may not be efficient in mining large sequence databases having numerous patterns and/or long patterns. In this paper, we propose a projection-based, sequential pattern-growth approach for efficient mining of sequential patterns. In this approach, a sequence database is recursively projected into a set of smaller projected databases, and sequential patterns are grown in each projected database by exploring only locally frequent fragments. Based on an initial study of the pattern growth-based sequential pattern mining, FreeSpan [J. Han et al. (2000)], we propose a more efficient method, called PSP, which offers ordered growth and reduced projected databases. To further improve the performance, a pseudoprojection technique is developed in PrefixSpan. A comprehensive performance study shows that PrefixSpan, in most cases, outperforms the a priori-based algorithm GSP, FreeSpan, and SPADE [M. Zaki, (2001)] (a sequential pattern mining algorithm that adopts vertical data format), and PrefixSpan integrated with pseudoprojection is the fastest among all the tested algorithms. Furthermore, this mining methodology can be extended to mining sequential patterns with user-specified constraints. The high promise of the pattern-growth approach may lead to its further extension toward efficient mining of other kinds of frequent patterns, such as frequent substructures.

Interactive sequence discovery by incremental mining

Sequential pattern mining has become a challenging task in data mining due to its complexity. Essentially, the mining algorithms discover all the frequent patterns meeting the user specified minimum support threshold. However, it is very unlikely that the user could obtain the satisfactory patterns in just one query. Usually the user must try various support thresholds to mine the database for the final desirable set of patterns. Consequently, the time-consuming mining process has to be repeated several times. However, current approaches are inadequate for such interactive mining due to the long processing time required for each query. In order to reduce the response time for each query during the interactive process, we propose a knowledge base assisted mining algorithm for interactive sequence discovery. The proposed approach utilizes the knowledge acquired from each mining process, accumulates the counting information to facilitate efficient counting of patterns, and speeds up the whole interactive mining process. Furthermore, the knowledge base makes possible the direct generation of new candidate sets and the concurrent support counting of variable sized candidates. Even for some queries, due to the pattern information already kept in the knowledge base, database access is not required at all. The conducted experiments show that our approach outperforms GSP, a state-of-the-art sequential pattern mining algorithm, by several order of magnitudes for interactive sequence discovery.

Mining hybrid sequential patterns and sequential rules

The problem addressed in this paper is to discover the frequently occurred sequential patterns from databases. Basically, the existing studies on finding sequential patterns can be roughly classified into two main categories. In the first category, the discovered patterns are continuous patterns, where all the elements in the pattern appear in consecutive positions in transactions. The second category is to mine discontinuous patterns, where the adjacent elements in the pattern need not appear consecutively in transactions. Although there are many researches on finding either kind of patterns, no previous researches can find both of them. Neither can they find the discontinuous patterns formed of several continuous sub-patterns. Therefore, we define a new kind of patterns, called hybrid pattern, which is the combination of continuous patterns and discontinuous patterns. In this paper, two algorithms are developed to mine hybrid patterns, where the first algorithm is easy but slow while the second complicated but much faster than the first one. Finally, the simulation result shows that our second algorithm is as fast as the currently best algorithm for mining sequential patterns.

A fast algorithm for mining sequential patterns from large databases

Mining sequential patterns from large databases has been recognized by many researchers as an attractive task of data mining and knowledge discovery. Previous algorithms scan the databases for many times, which is often unendurable due to the, very large amount of databases. In this paper, the authors introduce an effective algorithm for mining sequential patterns from large databases. In the algorithm, the original database is not used at all for counting the support of sequences after the first pass. Rather, a tidlist structure generated in the previous pass is employed for the purpose based on set intersection operations, avoiding the multiple scans of the databases.