Sliding window-based frequent pattern mining over data streams

Finding frequent patterns in a continuous stream of transactions is critical for many applications such as retail market data analysis, network monitoring, web usage mining, and stock market prediction. Even though numerous frequent pattern mining algorithms have been developed over the past decade, new solutions for handling stream data are still required due to the continuous, unbounded, and ordered sequence of data elements generated at a rapid rate in a data stream. Therefore, extracting frequent patterns from more recent data can enhance the analysis of stream data. In this paper, we propose an efficient technique to discover the complete set of recent frequent patterns from a high-speed data stream over a sliding window. We develop a Compact Pattern Stream tree (CPS-tree) to capture the recent stream data content and efficiently remove the obsolete, old stream data content. We also introduce the concept of dynamic tree restructuring in our CPS-tree to produce a highly compact frequency-descending tree structure at runtime. The complete set of recent frequent patterns is obtained from the CPS-tree of the current window using an FP-growth mining technique. Extensive experimental analyses show that our CPS-tree is highly efficient in terms of memory and time complexity when finding recent frequent patterns from a high-speed data stream.     

Mining frequent itemsets in data streams using the weighted sliding window model

In recent years, data stream mining has become an important research topic. With the emergence of new applications, the data we process are not again static, but the continuous dynamic data stream. Examples include network traffic analysis, Web click stream mining, network intrusion detection, and on-line transaction analysis. In this paper, we propose a new framework for data stream mining, called the weighted sliding window model. The proposed model allows the user to specify the number of windows for mining, the size of a window, and the weight for each window. Thus users can specify a higher weight to a more significant data section, which will make the mining result closer to user’s requirements. Based on the weighted sliding window model, we propose a single pass algorithm, called WSW, to efficiently discover all the frequent itemsets from data streams. By analyzing data characteristics, an improved algorithm, called WSW-Imp, is developed to further reduce the time of deciding whether a candidate itemset is frequent or not. Empirical results show that WSW-Imp outperforms WSW under the weighted sliding window model.     

数据流中一种基于滑动窗口的前K个频繁项集挖掘算法

数据流频繁项集挖掘是当今数据挖掘和知识学习领域重要的研究课题之一。数据流高速性、连续性、无界性、实时性对挖掘算法在时间和空间方面提出了更高的要求。传统的数据挖掘算法由于其存储结构需要频繁地维护,其挖掘方式的精度和速度较低,空间、时间效率不高。在基于粒计算和ECLAT算法的基础上提出一种挖掘数据流滑动窗口中top-K频繁项集算法,采用二进制方式存储项,利用位移运算实现增量更新,实施与运算计算项集支持度,同时利用二分查找法插入到项目序表中,输出前K个频繁项。实验结果表明,该算法在K取值不太高时具有较好的时空高效性。     

Approximate mining of maximal frequent itemsets in data streams with different window models

A data stream is a massive, open-ended sequence of data elements continuously generated at a rapid rate. Mining data streams is more difficult than mining static databases because the huge, high-speed and continuous characteristics of streaming data. In this paper, we propose a new one-pass algorithm called DSM-MFI (stands for Data Stream Mining for Maximal Frequent Itemsets), which mines the set of all maximal frequent itemsets in landmark windows over data streams. A new summary data structure called summary frequent itemset forest (abbreviated as SFI-forest) is developed for incremental maintaining the essential information about maximal frequent itemsets embedded in the stream so far. Theoretical analysis and experimental studies show that the proposed algorithm is efficient and scalable for mining the set of all maximal frequent itemsets over the entire history of the data streams.     

一种基于变尺度滑动窗口的数据流频繁集挖掘算法

基干传统滑动窗口机制的数据流频繁集挖掘算法较多地考虑快速且精确的效果,而较少考虑数据流的时变特性,对传统的滑动窗口机制进行改进.同时考虑数据流的海量特性和时变特性,提出一种基于变尺度滑动窗口机制的数据流频繁集挖掘算法V-Stream.该算法采用事务链表组的概要数据结构.能够根据数据流的数据分布变化自适应调整窗口大小.Eclipse上的仿真实验结果表明,V-Stream相比Manku算法提高了挖掘数据流频繁集的时间与空间效率.     

Towards a variable size sliding window model for frequent itemset mining over data streams

Sliding window is a widely used model for data stream mining due to its emphasis on recent data and its bounded memory requirement. The main idea behind a transactional sliding window is to keep a fixed size window over a data stream. The window size is kept constant by removing old transactions from the window, when new transactions arrive. Older transactions of window are removed irrespective to whether a significant change has occurred or not. Another challenge of sliding window model is determining window size. The classic approach for determining the window size is to obtain it from the user. In order to determine the precise size of the window, the user must have prior knowledge about the time and scale of changes within the data stream. However, due to the unpredictable changing nature of data streams, this prior knowledge cannot be easily determined. Moreover, by using a fixed window size during a data stream mining, the performance of this model is degraded in terms of reflecting recent changes. Based on these observations, this study relaxes the notion of window size and proposes a new algorithm named VSW (Variable Size sliding Window frequent itemset mining) which is suitable for observing recent changes in the set of frequent itemsets over data streams. The window size is determined dynamically based on amounts of concept change that occurs within the arriving data stream. The window expands as the concept becomes stable and shrinks when a concept change occurs. In this study, it is shown that if stale transactions are removed from the window after a concept change, updated frequent itemsets always belong to the most recent concept. Experimental evaluations on both synthetic and real data show that our algorithm effectively detects the concept change, adjust the window size, and adapts itself to the new concepts along the data stream.     

Maintaining frequent closed itemsets over a sliding window

In this paper, we study the incremental update of Frequent Closed Itemsets (FCIs) over a sliding window in a high-speed data stream. We propose the notion of semi-FCIs, which is to progressively increase the minimum support threshold for an itemset as it is retained longer in the window, thereby drastically reducing the number of itemsets that need to be maintained and processed. We explore the properties of semi-FCIs and observe that a majority of the subsets of a semi-FCI are not semi-FCIs and need not be updated. This finding allows us to devise an efficient algorithm, IncMine, that incrementally updates the set of semi-FCIs over a sliding window. We also develop an inverted index to facilitate the update process. Our empirical results show that IncMine achieves significantly higher throughput and consumes less memory than the state-of-the-art streaming algorithms for mining FCIs and FIs. IncMine also attains high accuracy of 100% precision and over 93% recall.     

基于FP-tree的最大频繁项目集挖掘算法

最大频繁项目集挖掘是数据挖掘领域最重要的基本问题之一,在分析已有算法的基础上提出了FP-MMFI算法,它是对FP-growth算法在最大频繁项目集挖掘上的扩展.提出了频繁路径的概念,用它可以有效地对FP-tree进行压缩和缩小搜索空间,同时使用投影的方法对超集检测进行了优化,减少了项目匹配的次数.最后实验结果表明,该算法在性能上优于已有的同类算法.     

Mining top-K frequent itemsets from data streams

Frequent pattern mining on data streams is of interest recently. However, it is not easy for users to determine a proper frequency threshold. It is more reasonable to ask users to set a bound on the result size. We study the problem of mining top K frequent itemsets in data streams. We introduce a method based on the Chernoff bound with a guarantee of the output quality and also a bound on the memory usage. We also propose an algorithm based on the Lossy Counting Algorithm. In most of the experiments of the two proposed algorithms, we obtain perfect solutions and the memory space occupied by our algorithms is very small. Besides, we also propose the adapted approach of these two algorithms in order to handle the case when we are interested in mining the data in a sliding window. The experiments show that the results are accurate.

Catch the moment: maintaining closed frequent itemsets over a data stream sliding window

This paper considers the problem of mining closed frequent itemsets over a data stream sliding window using limited memory space. We design a synopsis data structure to monitor transactions in the sliding window so that we can output the current closed frequent itemsets at any time. Due to time and memory constraints, the synopsis data structure cannot monitor all possible itemsets. However, monitoring only frequent itemsets will make it impossible to detect new itemsets when they become frequent. In this paper, we introduce a compact data structure, the closed enumeration tree (CET), to maintain a dynamically selected set of itemsets over a sliding window. The selected itemsets contain a boundary between closed frequent itemsets and the rest of the itemsets. Concept drifts in a data stream are reflected by boundary movements in the CET. In other words, a status change of any itemset (e.g., from non-frequent to frequent) must occur through the boundary. Because the boundary is relatively stable, the cost of mining closed frequent itemsets over a sliding window is dramatically reduced to that of mining transactions that can possibly cause boundary movements in the CET. Our experiments show that our algorithm performs much better than representative algorithms for the sate-of-the-art approaches. Yun Chi is currently a Ph.D. student at the Department of Computer Science, UCLA. His main areas of research include database systems, data mining, and bioinformatics. For data mining, he is interested in mining labeled trees and graphs, mining data streams, and mining data with uncertainty. Haixun Wang is currently a research staff member at IBM T. J. Watson Research Center. He received the B.S. and the M.S. degree, both in computer science, from Shanghai Jiao Tong University in 1994 and 1996. He received the Ph.D. degree in computer science from the University of California, Los Angeles in 2000. He has published more than 60 research papers in referred international journals and conference proceedings. He is a member of the ACM, the ACM SIGMOD, the ACM SIGKDD, and the IEEE Computer Society. He has served in program committees of international conferences and workshops, and has been a reviewer for some leading academic journals in the database field. Philip S. Yureceived the B.S. Degree in electrical engineering from National Taiwan University, the M.S. and Ph.D. degrees in electrical engineering from Stanford University, and the M.B.A. degree from New York University. He is with the IBM Thomas J. Watson Research Center and currently manager of the Software Tools and Techniques group. His research interests include data mining, Internet applications and technologies, database systems, multimedia systems, parallel and distributed processing, and performance modeling. Dr. Yu has published more than 430 papers in refereed journals and conferences. He holds or has applied for more than 250 US patents.Dr. Yu is a Fellow of the ACM and a Fellow of the IEEE. He is associate editors of ACM Transactions on the Internet Technology and ACM Transactions on Knowledge Discovery in Data. He is a member of the IEEE Data Engineering steering committee and is also on the steering committee of IEEE Conference on Data Mining. He was the Editor-in-Chief of IEEE Transactions on Knowledge and Data Engineering (2001–2004), an editor, advisory board member and also a guest co-editor of the special issue on mining of databases. He had also served as an associate editor of Knowledge and Information Systems. In addition to serving as program committee member on various conferences, he will be serving as the general chairman of 2006 ACM Conference on Information and Knowledge Management and the program chairman of the 2006 joint conferences of the 8th IEEE Conference on E-Commerce Technology (CEC' 06) and the 3rd IEEE Conference on Enterprise Computing, E-Commerce and E-Services (EEE' 06). He was the program chairman or co-chairs of the 11th IEEE International Conference on Data Engineering, the 6th Pacific Area Conference on Knowledge Discovery and Data Mining, the 9th ACM SIGMOD Workshop on Research Issues in Data Mining and Knowledge Discovery, the 2nd IEEE International Workshop on Research Issues on Data Engineering:Transaction and Query Processing, the PAKDD Workshop on Knowledge Discovery from Advanced Databases, and the 2nd IEEE International Workshop on Advanced Issues of E-Commerce and Web-based Information Systems. He served as the general chairman of the 14th IEEE International Conference on Data Engineering and the general co-chairman of the 2nd IEEE International Conference on Data Mining. He has received several IBM honors including 2 IBM Outstanding Innovation Awards, an Outstanding Technical Achievement Award, 2 Research Division Awards and the 84th plateau of Invention Achievement Awards. He received an Outstanding Contributions Award from IEEE International Conference on Data Mining in 2003 and also an IEEE Region 1 Award for “promoting and perpetuating numerous new electrical engineering concepts" in 1999. Dr. Yu is an IBM Master Inventor. Richard R. Muntz is a Professor and past chairman of the Computer Science Department, School of Engineering and Applied Science, UCLA. His current research interests are sensor rich environments, multimedia storage servers and database systems, distributed and parallel database systems, spatial and scientific database systems, data mining, and computer performance evaluation. He is the author of over one hundred and fifty research papers.Dr. Muntz received the BEE from Pratt Institute in 1963, the MEE from New York University in 1966, and the Ph.D. in Electrical Engineering from Princeton University in 1969. He is a member of the Board of Directors for SIGMETRICS and past chairman of IFIP WG7.3 on performance evaluation. He was a member of the Corporate Technology Advisory Board at NCR/Teradata, a member of the Science Advisory Board of NASA's Center of Excellence in Space Data Information Systems, and a member of the Goddard Space Flight Center Visiting Committee on Information Technology. He recently chaired a National Research Council study on “The Intersection of Geospatial Information and IT” which was published in 2003. He was an associate editor for the Journal of the ACM from 1975 to 1980 and the Editor-in-Chief of ACM Computing Surveys from 1992 to 1995. He is a Fellow of the ACM and a Fellow of the IEEE.     

基于改进FP-tree的最大频繁项集挖掘算法

现有的最大频繁项集挖掘算法在挖掘过程中需要进行超集检测,基于FP-tree的算法需要递归的建立条件频繁模式树,挖掘效率不高.提出了一种基于改进FP-tree高效挖掘最大频繁项集的算法(MMFI).该算法修改了FP-tree结构并采用NBN策略,在挖掘过程中既不需要进行超集检测也不需要递归的建立条件频繁模式树.算法分析和实验结果表明,该算法是一种有效、快速的算法.     

A false negative approach to mining frequent itemsets from high speed transactional data streams

Mining frequent itemsets from transactional data streams is challenging due to the nature of the exponential explosion of itemsets and the limit memory space required for mining frequent itemsets. Given a domain of I unique items, the possible number of itemsets can be up to 2^I − 1. When the length of data streams approaches to a very large number N, the possibility of an itemset to be frequent becomes larger and difficult to track with limited memory. The existing studies on finding frequent items from high speed data streams are false-positive oriented. That is, they control memory consumption in the counting processes by an error parameter ?, and allow items with support below the specified minimum support s but above s − ? counted as frequent ones. However, such false-positive oriented approaches cannot be effectively applied to frequent itemsets mining for two reasons. First, false-positive items found increase the number of false-positive frequent itemsets exponentially. Second, minimization of the number of false-positive items found, by using a small ?, will make memory consumption large. Therefore, such approaches may make the problem computationally intractable with bounded memory consumption. In this paper, we developed algorithms that can effectively mine frequent item(set)s from high speed transactional data streams with a bound of memory consumption. Our algorithms are based on Chernoff bound in which we use a running error parameter to prune item(set)s and use a reliability parameter to control memory. While our algorithms are false-negative oriented, that is, certain frequent itemsets may not appear in the results, the number of false-negative itemsets can be controlled by a predefined parameter so that desired recall rate of frequent itemsets can be guaranteed. Our extensive experimental studies show that the proposed algorithms have high accuracy, require less memory, and consume less CPU time. They significantly outperform the existing false-positive algorithms.     

Online mining of frequent sets in data streams with error guarantee

For most data stream applications, the volume of data is too huge to be stored in permanent devices or to be thoroughly scanned more than once. It is hence recognized that approximate answers are usually sufficient, where a good approximation obtained in a timely manner is often better than the exact answer that is delayed beyond the window of opportunity. Unfortunately, this is not the case for mining frequent patterns over data streams where algorithms capable of online processing data streams do not conform strictly to a precise error guarantee. Since the quality of approximate answers is as important as their timely delivery, it is necessary to design algorithms to meet both criteria at the same time. In this paper, we propose an algorithm that allows online processing of streaming data and yet guaranteeing the support error of frequent patterns strictly within a user-specified threshold. Our theoretical and experimental studies show that our algorithm is an effective and reliable method for finding frequent sets in data stream environments when both constraints need to be satisfied.     

A novel hash-based approach for mining frequent itemsets over data streams requiring less memory space

In recent times, data are generated as a form of continuous data streams in many applications. Since handling data streams is necessary and discovering knowledge behind data streams can often yield substantial benefits, mining over data streams has become one of the most important issues. Many approaches for mining frequent itemsets over data streams have been proposed. These approaches often consist of two procedures including continuously maintaining synopses for data streams and finding frequent itemsets from the synopses. However, most of the approaches assume that the synopses of data streams can be saved in memory and ignore the fact that the information of the non-frequent itemsets kept in the synopses may cause memory utilization to be significantly degraded. In this paper, we consider compressing the information of all the itemsets into a structure with a fixed size using a hash-based technique. This hash-based approach skillfully summarizes the information of the whole data stream by using a hash table, provides a novel technique to estimate the support counts of the non-frequent itemsets, and keeps only the frequent itemsets for speeding up the mining process. Therefore, the goal of optimizing memory space utilization can be achieved. The correctness guarantee, error analysis, and parameter setting of this approach are presented and a series of experiments is performed to show the effectiveness and the efficiency of this approach.     

DSM-FI: an efficient algorithm for mining frequent itemsets in data streams

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.

基于双向搜索的最大频繁项目集挖掘算法

结合自底向上与自顶向下的搜索策略,提出一种快速发现最大频繁项目集的算法.该算法利用非频繁项目集对候选最大频繁项目集进行剪枝和降维,减少了候选最大频繁项目集的数量,缩小了搜索空间,提高了算法的效率.算法分析和实验表明,该算法是一种有效、快速的算法.     

一种有效的基于图遍历的加权序列模式挖掘算法

为解决加权遍历模式挖掘问题,概括了加权有向图的种类,提出一种边加权有向图与顶点加权有向图间的变换模型,并基于该模型提出一种基于图遍历的加权序列模式挖掘算法GTWSPMiner.该算法根据遍历模式中的项的连续性特点,采用一种加权前缀投影序列模式增长方法,将原挖掘序列数据库的任务分解成一组挖掘局部投影数据库的小任务.对比实验结果表明,该算法能快速有效地挖掘加权频繁遍历模式.     

Discovering frequent itemsets over transactional data streams through an efficient and stable approximate approach

A data stream is a massive and unbounded sequence of data elements that are continuously generated at a fast speed. Compared with traditional approaches, data mining in data streams is more challenging since several extra requirements need to be satisfied. In this paper, we propose a mining algorithm for finding frequent itemsets over the transactional data stream. Unlike most of existing algorithms, our method works based on the theory of Approximate Inclusion–Exclusion. Without incrementally maintaining the overall synopsis of the stream, we can approximate the itemsets’ counts according to certain kept information and the counts bounding technique. Some additional techniques are designed and integrated into the algorithm for performance improvement. Besides, the performance of the proposed algorithm is tested and analyzed through a series of experiments.     

Incremental sequence-based frequent query pattern mining from XML queries

Existing algorithms of mining frequent XML query patterns (XQPs) employ a candidate generate-and-test strategy. They involve expensive candidate enumeration and costly tree-containment checking. Further, most of existing methods compute the frequencies of candidate query patterns from scratch periodically by checking the entire transaction database, which consists of XQPs transferred from user query logs. However, it is not straightforward to maintain such discovered frequent patterns in real XML databases as there may be frequent updates that may not only invalidate some existing frequent query patterns but also generate some new frequent query patterns. Therefore, a drawback of existing methods is that they are rather inefficient for the evolution of transaction databases. To address above-mentioned problems, this paper proposes an efficient algorithm ESPRIT to mine frequent XQPs without costly tree-containment checking. ESPRIT transforms XML queries into sequences using a one-to-one mapping technique and mines the frequent sequences to generate frequent XQPs. We propose two efficient incremental algorithms, ESPRIT-i and ESPRIT-i ⁺, to incrementally mine frequent XQPs. We devise several novel optimization techniques of query rewriting, cache lookup, and cache replacement to improve the answerability and the hit rate of caching. We have implemented our algorithms and conducted a set of experimental studies on various datasets. The experimental results demonstrate that our algorithms achieve high efficiency and scalability and outperform state-of-the-art methods significantly.     

基于最大频繁项目序列集挖掘ISS_DM算法的改进

阐述了挖掘最大频繁项目序列集ISS_DM算法,针对该算法不能有效地解决客户序列视图数据库的数据挖掘问题,结合序列模式提出了改进的ISS_DM算法,并进行了相应的验证.实践证明,改进后的算法同原算法相比,对相同的数据量进行挖掘,算法执行时间明显减少,效益较高.