数据挖掘的容错性研究

应用容错式挖掘,更多有用的信息可以从真实世界被噪音干扰的数据中获取.然而在过去相关的研究中,容错式频繁模式挖掘问题不是没有成熟的定义,就是将问题局限于仅找出允许固定错误数据项数的频繁模式.本文提出了比例性容错频繁模式挖掘的问题,并给出了两个实现算法:第一个算法称为FT-BottomUp,采用FT-Apriori定理找出允许所有可能错误数据项数的容错频繁模式;第二个算法FT-LevelWise,将所有的容错频繁模式按照它们所允许的错误数据项目数量分成多个群组,并分别挖掘出各个群组内的频繁模式.     

一种新的容错式频繁模式挖掘算法

应用容错式挖掘,更多有用的信息可以从真实世界中被噪音干扰的数据中获取。然而在过去相关的研究中,容错式频繁模式挖掘问题不是没有成熟的定义,就是将问题局限于仅找出允许固定错误数据项数的频繁模式。本文提出了比例性容错频繁模式挖掘的问题,并给出了两个实现算法。第一个算法称为FT-BottomUp,采用FT-Apriori定理找出允许所有可能错误数据项数的容错频繁模式。第二个算法FT-LevelWise将所有的容错频繁模式按照它们所允许的错误数据项目数量分成多个群组,并分别挖掘出各个群组内的频繁模式。实验结果证明我们所提出的方法可以挖掘出更多潜在的频繁模式。     

Mining Frequent Patterns without Candidate Generation: A Frequent-Pattern Tree Approach

Mining frequent patterns in transaction databases, time-series databases, and many other kinds of databases has been studied popularly in data mining research. Most of the previous studies adopt an Apriori-like candidate set generation-and-test approach. However, candidate set generation is still costly, especially when there exist a large number of patterns and/or long patterns.In this study, we propose a novel frequent-pattern tree (FP-tree) structure, which is an extended prefix-tree structure for storing compressed, crucial information about frequent patterns, and develop an efficient FP-tree-based mining method, FP-growth, for mining the complete set of frequent patterns by pattern fragment growth. Efficiency of mining is achieved with three techniques: (1) a large database is compressed into a condensed, smaller data structure, FP-tree which avoids costly, repeated database scans, (2) our FP-tree-based mining adopts a pattern-fragment growth method to avoid the costly generation of a large number of candidate sets, and (3) a partitioning-based, divide-and-conquer method is used to decompose the mining task into a set of smaller tasks for mining confined patterns in conditional databases, which dramatically reduces the search space. Our performance study shows that the FP-growth method is efficient and scalable for mining both long and short frequent patterns, and is about an order of magnitude faster than the Apriori algorithm and also faster than some recently reported new frequent-pattern mining methods.     

基于最右扩展枚举的半结构化数据最大模式挖掘方法研究

本文以标记有序树作为半结构化数据的数据模型 ,研究了半结构化数据的树状最大频繁模式挖掘问题 .已有挖掘算法通常挖掘所有频繁模式 ,其中很多模式为其它模式的子模式 ,针对该问题 ,设计实现了一种最大模式挖掘算法 .该算法采用最右扩展枚举方法无重复枚举所有候选模式 ,利用频繁模式扩展森林实现高效剪枝扩展和挖掘频繁叶模式 ,通过计算频繁叶模式间的包含关系挖掘树状最大频繁模式 .试验结果表明该算法具有良好性能     

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.     

FP-growth算法的实现方法研究

事务数据库中频繁模式的挖掘研究作为关联规则等许多数据挖掘问题的核心工作,已经研究了许多年。早期算法大都是Apriori型算法,即首先产生候选集,然后在候选集的基础上找出频繁模式,候选集的产生往往是耗时的,特别是挖掘富模式或长模式时。JianweiHan等人提出了一种新颖的数据结构FP-tree及基于其上的FP-growth算法,用于有效的富模式与长模式挖掘。由于不同的实现方法可能会导致不同的挖掘效率,该文在讨论FP-growth算法的基础上,采用了几种不同的方法来实现它,并用几个数据库对它们的性能进行了比较。     

数据流中基于滑动窗口的序列模式挖掘算法

序列模式发现是最重要的数据挖掘任务之一,并有着广阔的应用前景。针对静态数据库,序列模式挖掘已经被深入地研究,但针对基于数据流的序列模式挖掘的研究还不是十分深入。数据流有着无限性的特性,因此往往不能保存数据流中全部的数据,同时很多时候只对最近的时间段的序列模式感兴趣,提出一个有效的结合滑动窗口技术的挖掘序列模式的算法FPM-SW,算法利用到3个数据结构（PatternTable,CountTable和Ta-tree）来处理基于数据流的序列模式挖掘的复杂性问题。算法通过CountTable结构来保存以往的潜在频繁序列,考虑到在某些情况下CountTable占用内存过多,算法还结合了一种压缩CountTable技术来减少内存占用。FPM-SW的优点是可以最大限度地降低负正例的产生,实验表明FPM-SW具有较高的准确率。     

Mining top?k frequent patterns without minimum support threshold

Finding frequent patterns play an important role in mining association rules, sequences, episodes, Web log mining and many other interesting relationships among data. Frequent pattern mining methods often produce a huge number of frequent itemsets that is not feasible for effective usage. The number of highly correlated patterns is usually very small and may even be one. Most of the existing frequent pattern mining techniques often require the setting of many input parameters and may involve multiple passes over the database. Minimum support is the widely used parameter in frequent pattern mining to discover statistically significant patterns. Specifying appropriate minimum support is a challenging task for a data analyst as the choice of minimum support value is somewhat arbitrary. Generally, it is required to repeatedly execute an algorithm, heuristically tuning the value of minimum support over a wide range, until the desired result is obtained, certainly, a very time-consuming process. Setting up an inappropriate minimum support may also cause an algorithm to fail in finding the true patterns. We present a novel method to efficiently retrieve top few maximal frequent patterns in order of significance without use of the minimum support parameter. Instead, we are only required to specify a more human understandable parameter, namely the desired number itemsets k. Our technique requires only a single pass over the database and generation of length two itemsets. The association ratio graph is proposed as a compact structure containing concise information, which is created in time quadratic to the size of the database. Algorithms are described for using this graph structure to discover top-most and top-k maximal frequent itemsets without minimum support threshold. To effectively achieve this, the method employs construction of an all path source-to-destination tree to discover all maximal cycles in the graph. The results can be ranked in decreasing order of significance. Results are presented demonstrating the performance advantages to be gained from the use of this approach.     

Effective periodic pattern mining in time series databases

The goal of analyzing a time series database is to find whether and how frequent a periodic pattern is repeated within the series. Periodic pattern mining is the problem that regards temporal regularity. However, most of the existing algorithms have a major limitation in mining interesting patterns of users interest, that is, they can mine patterns of specific length with all the events sequentially one after another in exact positions within this pattern. Though there are certain scenarios where a pattern can be flexible, that is, it may be interesting and can be mined by neglecting any number of unimportant events in between important events with variable length of the pattern. Moreover, existing algorithms can detect only specific type of periodicity in various time series databases and require the interaction from user to determine periodicity. In this paper, we have proposed an algorithm for the periodic pattern mining in time series databases which does not rely on the user for the period value or period type of the pattern and can detect all types of periodic patterns at the same time, indeed these flexibilities are missing in existing algorithms. The proposed algorithm facilitates the user to generate different kinds of patterns by skipping intermediate events in a time series database and find out the periodicity of the patterns within the database. It is an improvement over the generating pattern using suffix tree, because suffix tree based algorithms have weakness in this particular area of pattern generation. Comparing with the existing algorithms, the proposed algorithm improves generating different kinds of interesting patterns and detects whether the generated pattern is periodic or not. We have tested the performance of our algorithm on both synthetic and real life data from different domains and found a large number of interesting event sequences which were missing in existing algorithms and the proposed algorithm was efficient enough in generating and detecting periodicity of flexible patterns on both types of data.     

一种直接在Trans-树中挖掘频繁模式的新算法

Frequent pattern mining plays an essential role in many important data mining tasks. FP-growth is a very efficient algorithm for frequent pattern mining. However, it still suffers from creating conditional FP-tree separately and recursively during the mining process. In this paper, we propose a new algorithm, called Least-Item-First Pat-tern Growth (LIFPG), for mining frequent patterns. LIFPG mines frequent patterns directly in Trans-tree withoutusing any additional data structures. The key idea is that least items are always considered first when the current pat-tern growth. By this way, conditional sub-tree can be created directly in Trans-tree by adjusting node-links and re-counting counts of some nodes. Experiments show that, in comparison with FP-Growth, our algorithm is about fourtimes faster and saves half of memory;it also has good time and space scalability with the number of transactions,and has an excellent performance in dense dataset mining as well.     

Sliding window based weighted maximal frequent pattern mining over data streams

As data have been accumulated more quickly in recent years, corresponding databases have also become huger, and thus, general frequent pattern mining methods have been faced with limitations that do not appropriately respond to the massive data. To overcome this problem, data mining researchers have studied methods which can conduct more efficient and immediate mining tasks by scanning databases only once. Thereafter, the sliding window model, which can perform mining operations focusing on recently accumulated parts over data streams, was proposed, and a variety of mining approaches related to this have been suggested. However, it is hard to mine all of the frequent patterns in the data stream environment since generated patterns are remarkably increased as data streams are continuously extended. Thus, methods for efficiently compressing generated patterns are needed in order to solve that problem. In addition, since not only support conditions but also weight constraints expressing items’ importance are one of the important factors in the pattern mining, we need to consider them in mining process. Motivated by these issues, we propose a novel algorithm, weighted maximal frequent pattern mining over data streams based on sliding window model (WMFP-SW) to obtain weighted maximal frequent patterns reflecting recent information over data streams. Performance experiments report that MWFP-SW outperforms previous algorithms in terms of runtime, memory usage, and scalability.     

带通配符和One-Off条件的序列模式挖掘

很多应用领域产生大量的序列数据。如何从这些序列数据中挖掘具有重要价值的模式,已成为序列模式挖掘研究的主要任务。研究这样一个问题：给定序列S、支持度阈值和间隔约束,从序列S中挖掘所有出现次数不小于给定支持度阈值的频繁序列模式,并且要求模式中任意两个相邻元素在序列中的出现位置满足用户定义的间隔约束。设计了一种有效的带有通配符的模式挖掘算法One-Off Mining,模式在序列中的出现满足One-Off条件,即模式的任意两次出现都不共享序列中同一位置的字符。在生物DNA序列上的实验结果表明,One-Off Mining比相关的序列模式挖掘算法具有更好的时间性能和完备性。     

Efficient mining of high utility pattern with considering of rarity and length

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.     

Mining globally interesting patterns from multiple databases using kernel estimation

When extracting knowledge (or patterns) from multiple databases, the data from different databases might be too large in volume to be merged into one database for centralized mining on one computer, the local information sources might be hidden from a global decision maker due to privacy concerns, and different local databases may have different contribution to the global pattern. Dealing with multiple databases is essentially different from mining from a single database. In multi-database mining, the global patterns must be obtained by carefully analyzing the local patterns from individual databases. In this paper, we propose a nonlinear method, named KEMGP (kernel estimation for mining global patterns), to tackle this problem, which adopts kernel estimation to synthesizing local patterns for global patterns. We also adopt a method to divide all the data in different databases according to attribute dimensionality, which reduces the total space complexity. We test our algorithm on a customer management system, where the application is to obtain all globally interesting patterns by analyzing the individual databases. The experimental results show that our method is efficient.     

Closed inter-sequence pattern mining

Inter-sequence pattern mining can find associations across several sequences in a sequence database, which can discover both a sequential pattern within a transaction and sequential patterns across several different transactions. However, inter-sequence pattern mining algorithms usually generate a large number of recurrent frequent patterns. We have observed mining closed inter-sequence patterns instead of frequent ones can lead to a more compact yet complete result set. Therefore, in this paper, we propose a model of closed inter-sequence pattern mining and an efficient algorithm called CISP-Miner for mining such patterns, which enumerates closed inter-sequence patterns recursively along a search tree in a depth-first search manner. In addition, several effective pruning strategies and closure checking schemes are designed to reduce the search space and thus accelerate the algorithm. Our experiment results demonstrate that the proposed CISP-Miner algorithm is very efficient and outperforms a compared EISP-Miner algorithm in most cases.     

Sequential Pattern Mining in Multi-Databases via Multiple Alignment

To efficiently find global patterns from a multi-database, information in each local database must first be mined and summarized at the local level. Then only the summarized information is forwarded to the global mining process. However, conventional sequential pattern mining methods based on support cannot summarize the local information and is ineffective for global pattern mining from multiple data sources. In this paper, we present an alternative local mining approach for finding sequential patterns in the local databases of a multi-database. We propose the theme of approximate sequential pattern mining roughly defined as identifying patterns approximately shared by many sequences. Approximate sequential patterns can effectively summerize and represent the local databases by identifying the underlying trends in the data. We present a novel algorithm, ApproxMAP, to mine approximate sequential patterns, called consensus patterns, from large sequence databases in two steps. First, sequences are clustered by similarity. Then, consensus patterns are mined directly from each cluster through multiple alignment. We conduct an extensive and systematic performance study over synthetic and real data. The results demonstrate that ApproxMAP is effective and scalable in mining large sequences databases with long patterns. Hence, ApproxMAP can efficiently summarize a local database and reduce the cost for global mining. Furthremore, we present an elegant and uniform model to identify both high vote sequential patterns and exceptional sequential patterns from the collection of these consensus patterns from each local databases.     

Tree pattern mining with tree automata constraints

Most work on pattern mining focuses on simple data structures such as itemsets and sequences of itemsets. However, a lot of recent applications dealing with complex data like chemical compounds, protein structures, XML and Web log databases and social networks, require much more sophisticated data structures such as trees and graphs. In these contexts, interesting patterns involve not only frequent object values (labels) appearing in the graphs (or trees) but also frequent specific topologies found in these structures. Recently, several techniques for tree and graph mining have been proposed in the literature. In this paper, we focus on constraint-based tree pattern mining. We propose to use tree automata as a mechanism to specify user constraints over tree patterns. We present the algorithm CoBMiner which allows user constraints specified by a tree automata to be incorporated in the mining process. An extensive set of experiments executed over synthetic and real data (XML documents and Web usage logs) allows us to conclude that incorporating constraints during the mining process is far more effective than filtering the interesting patterns after the mining process.     

Mining closed and maximal frequent subtrees from databases of labeled rooted trees

Tree structures are used extensively in domains such as computational biology, pattern recognition, XML databases, computer networks, and so on. One important problem in mining databases of trees is to find frequently occurring subtrees. Because of the combinatorial explosion, the number of frequent subtrees usually grows exponentially with the size of frequent subtrees and, therefore, mining all frequent subtrees becomes infeasible for large tree sizes. We present CMTreeMiner, a computationally efficient algorithm that discovers only closed and maximal frequent subtrees in a database of labeled rooted trees, where the rooted trees can be either ordered or unordered. The algorithm mines both closed and maximal frequent subtrees by traversing an enumeration tree that systematically enumerates all frequent subtrees. Several techniques are proposed to prune the branches of the enumeration tree that do not correspond to closed or maximal frequent subtrees. Heuristic techniques are used to arrange the order of computation so that relatively expensive computation is avoided as much as possible. We study the performance of our algorithm through extensive experiments, using both synthetic data and data sets from real applications. The experimental results show that our algorithm is very efficient in reducing the search space and quickly discovers all closed and maximal frequent subtrees.     

Discovering Transitional Patterns and Their Significant Milestones in Transaction Databases

A transaction database usually consists of a set of time-stamped transactions. Mining frequent patterns in transaction databases has been studied extensively in data mining research. However, most of the existing frequent pattern mining algorithms (such as Apriori and FP-growth) do not consider the time stamps associated with the transactions. In this paper, we extend the existing frequent pattern mining framework to take into account the time stamp of each transaction and discover patterns whose frequency dramatically changes over time. We define a new type of patterns, called transitional patterns, to capture the dynamic behavior of frequent patterns in a transaction database. Transitional patterns include both positive and negative transitional patterns. Their frequencies increase/decrease dramatically at some time points of a transaction database. We introduce the concept of significant milestones for a transitional pattern, which are time points at which the frequency of the pattern changes most significantly. Moreover, we develop an algorithm to mine from a transaction database the set of transitional patterns along with their significant milestones. Our experimental studies on real-world databases illustrate that mining positive and negative transitional patterns is highly promising as a practical and useful approach for discovering novel and interesting knowledge from large databases.