Data Stream Mining Research Articles

An Efficient Frequent Itemset Mining Method over High-speed Data Streams

Frequent itemset mining over sliding window is an interesting problem and has a large number of applications. Sliding window is a widely used model for frequent itemset mining over data streams due to its capability to handle concept drift, its bounded memory and its low processing time. A sliding window-based algorithm requires an efficient data structure that can be updated as fast as possible when inserting and deleting transactions. Moreover, an innovative computing method is needed to provide the set of frequent patterns (FPs) with a little delay after the user issues a request for the mining results within a window. In this study, an efficient representation of the sliding window named blocked bit sequence is introduced which is aimed to store and maintain the content of the window. Moreover, by a novel technique this representation is exploited for efficiently extracting the set of FPs within the current window. Experimental evaluations on both real-life and synthetic data streams show that the proposed approach is faster than recently proposed algorithms in different phases of data stream mining.

An Index-Inspired Algorithm for Anytime Classification on Evolving Data Streams

Due to the ever growing presence of data streams there has been a considerable amount of research on stream data mining over the past years. Anytime algorithms are particularly well suited for stream mining, since they flexibly use all available time on streams of varying data rates, and are also shown to outperform traditional budget approaches on constant streams. In this article we present an index-inspired algorithm for Bayesian anytime classification on evolving data streams and show its performance on benchmark data sets.

Single-pass incremental and interactive mining for weighted frequent patterns

Weighted frequent pattern (WFP) mining is more practical than frequent pattern mining because it can consider different semantic significance (weight) of the items. For this reason, WFP mining becomes an important research issue in data mining and knowledge discovery. However, existing algorithms cannot be applied for incremental and interactive WFP mining and also for stream data mining because they are based on a static database and require multiple database scans. In this paper, we present two novel tree structures IWFPTWA (Incremental WFP tree based on weight ascending order) and IWFPTFD (Incremental WFP tree based on frequency descending order), and two new algorithms IWFPWA and IWFPFD for incremental and interactive WFP mining using a single database scan. They are effective for incremental and interactive mining to utilize the current tree structure and to use the previous mining results when a database is updated or a minimum support threshold is changed. IWFPWA gets advantage in candidate pattern generation by obtaining the highest weighted item in the bottom of IWFPTWA. IWFPFD ensures that any non-candidate item cannot appear before candidate items in any branch of IWFPTFD and thus speeds up the prefix tree and conditional tree creation time during mining operation. IWFPTFD also achieves the highly compact incremental tree to save memory space. To our knowledge, this is the first research work to perform single-pass incremental and interactive mining for weighted frequent patterns. Extensive performance analyses show that our tree structures and algorithms are very efficient and scalable for single-pass incremental and interactive WFP mining.

Positive and negative association rule mining on XML data streams in database as a service concept

In recent years, data mining has become one of the most popular techniques for data owners to determine their strategies. Association rule mining is a data mining approach that is used widely in traditional databases and usually to find the positive association rules. However, there are some other challenging rule mining topics like data stream mining and negative association rule mining. Besides, organizations want to concentrate on their own business and outsource the rest of their work. This approach is named “database as a service concept” and provides lots of benefits to data owner, but, at the same time, brings out some security problems. In this paper, a rule mining system has been proposed that provides efficient and secure solution to positive and negative association rule computation on XML data streams in database as a service concept. The system is implemented and several experiments have been done with different synthetic data sets to show the performance and efficiency of the proposed system.

Data stream classification with artificial endocrine system

Due to concept drifts, maintaining an up-to-date model is a challenging task for most of the current classification approaches used in data stream mining. Both the incremental classifiers and the e...

A Novel Strategy for Mining Frequent Closed Itemsets in Data Streams

Mining frequent itemsets from data stream is an important task in stream data mining. This paper presents an algorithm Stream_FCI for mining the frequent closed itemsets from data streams in the model of sliding window. The algorithm detects the frequent closed itemsets in each sliding window using a DFP-tree with a head table. In processing each new transaction, the algorithm changes the head table and modifies the DFP-tree according to the changed items in the head table. The algorithm also adopts a table to store the frequent closed itemsets so as to avoid the time-consuming operations of searching in the whole DFP-tree for adding or deleting transactions. Our experimental results show that our algorithm is more efficient and has lower time and memory complexity than the similar algorithms Moment and FPCFI-DS.

Real-time clinical decision support system with data stream mining.

This research aims to describe a new design of data stream mining system that can analyze medical data stream and make real-time prediction. The motivation of the research is due to a growing concern of combining software technology and medical functions for the development of software application that can be used in medical field of chronic disease prognosis and diagnosis, children healthcare, diabetes diagnosis, and so forth. Most of the existing software technologies are case-based data mining systems. They only can analyze finite and structured data set and can only work well in their early years and can hardly meet today's medical requirement. In this paper, we describe a clinical-support-system based data stream mining technology; the design has taken into account all the shortcomings of the existing clinical support systems.

Gender Identification on Twitter Using the Modified Balanced Winnow

With the rapid growth of web-based social networking technologies in recent years, author identification and analysis have proven increasingly useful. Authorship analysis provides information about a document’s author, often including the author’s gender. Men and women are known to write in distinctly different ways, and these differences can be successfully used to make a gender prediction. Making use of these distinctions between male and female authors, this study demonstrates the use of a simple stream-based neural network to automatically discriminate gender on manually labeled tweets from the Twitter social network. This neural network, the Modified Balanced Winnow, was employed in two ways; the effectiveness of data stream mining was initially examined with an extensive list of n-gram features. Feature selection techniques were then evaluated by drastically reducing the feature list using WEKA’s attribute selection algorithms. This study demonstrates the effectiveness of the stream mining approach, achieving an accuracy of 82.48%, a 20.81% increase above the baseline prediction. Using feature selection methods improved the results by an additional 16.03%, to an accuracy of 98.51%.

Overview Of Streaming-Data Algorithms

Due to recent advances in data collection techniques, massive amounts of data are being collected at an extremely fast pace. Also, these data are potentially unbounded. Boundless streams of data collected from sensors, equipments, and other data sources are referred to as data streams. Various data mining tasks can be performed on data streams in search of interesting patterns. This paper studies a particular data mining task, clustering, which can be used as the first step in many knowledge discovery processes. By grouping data streams into homogeneous clusters, data miners can learn about data characteristics which can then be developed into classification models for new data or predictive models for unknown events. Recent research addresses the problem of data-stream mining to deal with applications that require processing huge amounts of data such as sensor data analysis and financial applications. For such analysis, single-pass algorithms that consume a small amount of memory are critical.

Compression and aggregation of Bayesian estimates for data intensive computing

Bayesian estimation is a major and robust estimator for many advanced statistical models. Being able to incorporate prior knowledge in statistical inference, Bayesian methods have been successfully applied in many different fields such as business, computer science, economics, epidemiology, genetics, imaging, and political science. However, due to its high computational complexity, Bayesian estimation has been deemed difficult, if not impractical, for large-scale databases, stream data, data warehouses, and data in the cloud. In this paper, we propose a novel compression and aggregation schemes (C&A) that enables distributed, parallel, or incremental computation of Bayesian estimates. Assuming partitioning of a large dataset, the C&A scheme compresses each partition into a synopsis and aggregates the synopsis into an overall Bayesian estimate without accessing the raw data. Such a C&A scheme can find applications in OLAP for data cubes, stream data mining, and cloud computing. It saves tremendous computing time since it processes each partition only once, enabling fast incremental update, and allows parallel processing. We prove that the compression is asymptotically lossless in the sense that the aggregated estimator deviates from the true model by an error that is bounded and approaches to zero when the data size increases. The results show that the proposed C&A scheme can make feasible OLAP of Bayesian estimates in a data cube. Further, it supports real-time Bayesian analysis of stream data, which can only be scanned once and cannot be permanently retained. Experimental results validate our theoretical analysis and demonstrate that our method can dramatically save time and space costs with almost no degradation of the modeling accuracy.

Advances in data stream mining

AbstractMining data streams has been a focal point of research interest over the past decade. Hardware and software advances have contributed to the significance of this area of research by introducing faster than ever data generation. This rapidly generated data has been termed as data streams. Credit card transactions, Google searches, phone calls in a city, and many others\are typical data streams. In many important applications, it is inevitable to analyze this streaming data in real time. Traditional data mining techniques have fallen short in addressing the needs of data stream mining. Randomization, approximation, and adaptation have been used extensively in developing new techniques or adopting exiting ones to enable them to operate in a streaming environment. This paper reviews key milestones and state of the art in the data stream mining area. Future insights are also be presented. © 2011 Wiley Periodicals, Inc.This article is categorized under: Algorithmic Development > Spatial and Temporal Data Mining

Knowledge discovery in data streams with regression tree methods

AbstractThis paper presents an advanced review of regression tree methods for mining data streams. Batch regression tree methods are known for their simplicity, interpretability, accuracy, and efficiency. They use fast divide‐and‐conquer greedy algorithms that recursively partition the given training data into smaller subsets. The result is a tree‐shaped model with splitting rules in the internal nodes and predictions in the leaves. Most batch regression tree methods take a complete dataset and build a model using that data. Generally, this tree model cannot be modified if new data is acquired later. Their successors, the incremental model and interval trees algorithms, are able to build and retrain a model on a step‐by‐step basis by incorporating new numerical training instances into the model as they become available. Moreover, these algorithms produce even more compact and accurate models than batch regression tree algorithms because they use intervals or functional models with a change detection mechanism, which makes them a more suitable choice for regression analysis of data streams. Finally, this review summarizes the performance results of the reviewed methods and crystallizes 10 requirements for successful implementation of a regression tree algorithm in data stream mining area. © 2011 Wiley Periodicals, Inc.This article is categorized under: Algorithmic Development > Spatial and Temporal Data Mining Technologies > Machine Learning Technologies > Prediction

데이터 스트림 마이닝에서 정보 중요성 차별화를 위한 퍼지 윈도우 기법

Considering the characteristics of a data stream whose data elements are continuously generated and may change over time, there have been many techniques to differentiate the importance of data elements in a data stream by their generation time. The conventional techniques are efficient to get an analysis result focusing on the recent information in a data stream, but they have a limitation to differentiate the importance of information in various ways more flexible. An information differentiation technique based on the term of a fuzzy set can be an alternative way to compensate the limitation. A term of a fuzzy set has been widely used in various data mining fields, which can overcome the sharp boundary problem and give an analysis result reflecting the requirements in real world applications more. In this paper, a fuzzy window mechanism is proposed, which is adapting a term of a fuzzy set and is efficiently used to differentiate the importance of information in mining data streams. Basic concepts including fuzzy calendars are described first, and subsequently details on data stream mining of weighted patterns using a fuzzy window technique are described.

MHUI-max: An efficient algorithm for discovering high-utility itemsets from data streams

Online mining of utility itemsets from data streams is one of the most interesting research issues in stream data mining. Although a number of relevant approaches have been proposed in recent years, they have the drawback of producing a large number of candidate itemsets for high-utility itemset mining. In this paper, an efficient algorithm, called MHUI-max (Mining High-Utility Itemsets based on LexTree-maxHTU), is proposed for mining high-utility itemsets from data streams with fewer candidates. Based on the framework of the MHUI-max algorithm, an effective representation of item information, called TID-list, and a new lexicographical tree-based data structure, called LexTree-maxHTU, has been developed to improve the efficiency of discovering high-utility itemsets with positive profits from data streams. Experimental results show that the proposed algorithm, MHUI-max, outperforms the existing approaches, MHUI-TID and THUI-Mine, for mining high-utility itemsets from data streams over transaction-sensitive sliding windows.

A Precise Statistical approach for concept change detection in unlabeled data streams

Recently data stream has been extensively explored due to its emergence in a great deal of applications such as sensor networks, web click streams and network flows. One of the most important challenges in data streams is concept change where data underlying distributions change from time to time. A vast majority of researches in the context of data stream mining are devoted to labeled data, whereas, in real word human practice label of data are rarely available to the learning algorithms. Moreover, most of the methods that detect changes in unlabeled data stream merely deal with numerical data sets, and also, they are facing considerable difficulty when dimension of data tends to increase. In this paper, we present a Precise Statistical approach for Concept Change Detection in unlabeled data streams, which, abbreviated as PSCCD, detects changes using an exchangeable test. This hypothesis test is driven from a martingale which is based on Doob’s Maximal Inequality. The advantages of our approach are three fold. First, it does not require a sliding window on the data stream whose size is a well-known challenging issue; second, it works well in multi-dimensional data stream, and last but not the least, it is applicable to different types of data including categorical, numerical and mixed-attribute data streams. To explore the advantages of our approach, quite a lot of experiments with different settings and specifications are conducted. The obtained results are very promising.

A Density-based Clustering over Evolving Heterogeneous Data Stream

Data stream clustering is an importance issue in data stream mining. In most of the existing algorithms, only the continuous features are used for clustering. In this paper, we introduce an algorithm HDenStream for clustering data stream with heterogeneous features. The HDenstream is also a density-based algorithm, so it is capable enough to cluster arbitrary shapes and handle outliers. Theoretic analysis and experimental results show that HDenStream is effective and efficient.

Feature Extracting of Business Data Streams with Concept-drifting

Business data streams are dynamic and easy to drift, thus extracting concept-drifting feature is one important work of data streams mining. This paper describes the characteristics and the concept drift of data streams, proposes work flow of concept formal analysis and the formal concept description model of streaming data based on granular computing. Concept-drifting in business data streams is actually decided by the changes upon the extension of the concept. Then, the paper describes concept coincidence, including coincidence on extent, coincidence on intent and coincidence on concept. Because concept-lattices can be expressed in terms of the granulated data streams, we analyze concept lattice pairs’ coincidence instead of continued and non-formal streaming data. And then the paper proposes the concept lattice pairs’ based concept relaxation-matching coincidence degree algorithm; the feature extraction method is also described. Finally, experiment and analysis are presented in order to explain and evaluate the method.

Incremental Mining of Closed Sequential Patterns in Multiple Data Streams

Sequential pattern mining searches for the relative sequence of events, allowing users to make predictions on discovered sequential patterns. Due to drastically advanced information technology over recent years, data have rapidly changed, growth in data amount has exploded and real-time demand is increasing, leading to the data stream environment. Data in this environment cannot be fully stored and ineptitude in traditional mining techniques has led to the emergence of data stream mining technology. Multiple data streams are a branch of the data stream environment. The MILE algorithm cannot preserve previously mined sequential patterns when new data are entered because of the concept of one-time fashion mining. To address this problem, we propose the ICspan algorithm to continue mining sequential patterns through an incremental approach and to acquire a more accurate mining result. In addition, due to the algorithm constraint in closed sequential patterns mining, the generation and records for sequential patterns will be reduced, leading to a decrease of memory usage and to an effective increase of execution efficiency.

The Six Technical Gaps between Intelligent Applications and Real-Time Data Mining: A Critical Review

Intelligent application is characterized by its ability to make decision autonomously. Some examples are distributed agents that collaboratively complete a complex task, yet each one of them is able to work and reason independently given the dynamic situation where they are in. The underlying think tank is often a collection of core mechanisms that include environment sensing, data capturing, data mining, ETL, knowledge processing, and decision making etc. All these techniques when placed and function together as a whole intelligent system, they will have to fulfill stringent deadlines imposed by the requirements of a real-time system. In the literature many research papers can be found on a wide variety of data mining techniques that enable intelligent applications operating in real-time. This report offers a critical review of the relevant literature, and contributes to the knowledge of identifying their shortcomings, so-called the technical gaps between the real-time requirements of a general intelligent system and the supporting components. A discussion follows on the possibilities of future intelligent applications that are empowered by data stream mining.

ApproxCCA: An approximate correlation analysis algorithm for multidimensional data streams

Correlation analysis is regarded as a significant challenge in the mining of multidimensional data streams. Great emphasis is generally placed on one-dimensional data streams with the existing correlation analysis methods for the mining of data streams. Therefore, the identification of underlying correlation among multivariate arrays (e.g. Sensor data) has long been ignored. The technique of canonical correlation analysis (CCA) has rarely been applied in multidimensional data streams. In this study, a novel correlation analysis algorithm based on CCA, called ApproxCCA, is proposed to explore the correlations between two multidimensional data streams in the environment with limited resources. By introducing techniques of unequal probability sampling and low-rank approximation to reduce the dimensionality of the product matrix composed by the sample covariance matrix and sample variance matrix, ApproxCCA successfully improves computational efficiency while ensuring the analytical precision. Experimental results of synthetic and real data sets have indicated that the computational bottleneck of traditional CCA can be overcome with ApproxCCA, and the correlations between two multidimensional data streams can also be detected accurately.