Because of the increasing data sizes and demands for low latency in modern data analysis, the traditional data warehousing technologies are greatly pushed beyond their limits. Several stream data warehouse (SDW) systems, which are warehouses that ingest append-only data feeds and support frequent refresh cycles, have been proposed including different methods to improve the responsiveness of the systems. Materialized views are critical in large-scale data warehouses due to their ability to speed up queries. Thus an SDW maintains layers of materialized views. Materialized view maintenance in SDW systems introduces new challenges. However, some of the existing SDW systems do not address the maintenance of views while others employ view maintenance techniques that are not efficient. This thesis presents ViewDF, a flexible framework for incremental maintenance of materialized views in SDW systems that generalizes existing techniques and enables new optimizations for views defined with operators that are common in stream analytics. We give a special view definition (ViewDF) to enhance the traditional way of creating views in SQL by being able to reference any partition of any table. We describe a prototype system based on this idea, which allows users to write ViewDFs directly and can automatically translate a broad class of queries into ViewDFs. Several optimizations are proposed and experiments show that our proposed system can improve view maintenance time by a factor of two or more in practical settings.

No description available.

This thesis explores the data modeling for outlier detection techniques in three different application domains: maritime surveillance, district heating, and online media and sequence datasets. The proposed models are evaluated and validated under different experimental scenarios, taking into account specific characteristics and setups of the different domains. Outlier detection has been studied and applied in many domains. Outliers arise due to different reasons such as fraudulent activities, structural defects, health problems, and mechanical issues. The detection of outliers is a challenging task that can reveal system faults, fraud, and save people's lives. Outlier detection techniques are often domain-specific. The main challenge in outlier detection relates to modeling the normal behavior in order to identify abnormalities. The choice of model is important, i.e., an incorrect choice of data model can lead to poor results. This requires a good understanding and interpretation of the data, the constraints, and the requirements of the problem domain. Outlier detection is largely an unsupervised problem due to unavailability of labeled data and the fact that labeled data is expensive. We have studied and applied a combination of both machine learning and data mining techniques to build data-driven and domain-oriented outlier detection models. We have shown the importance of data preprocessing as well as feature selection in building suitable methods for data modeling. We have taken advantage of both supervised and unsupervised techniques to create hybrid methods. For example, we have proposed a rule-based outlier detection system based on open data for the maritime surveillance domain. Furthermore, we have combined cluster analysis and regression to identify manual changes in the heating systems at the building level. Sequential pattern mining for identifying contextual and collective outliers in online media data have also been exploited. In addition, we have proposed a minimum spanning tree clustering technique for detection of groups of outliers in online media and sequence data. The proposed models have been shown to be capable of explaining the underlying properties of the detected outliers. This can facilitate domain experts in narrowing down the scope of analysis and understanding the reasons of such anomalous behaviors. We have also investigated the reproducibility of the proposed models in similar application domains. / Scalable resource-efficient systems for big data analytics

Big Data analytics has attracted intense interest recently for its attempt to extract information, knowledge and wisdom from Big Data. In industry, with the development of sensor technology and Information & Communication Technologies (ICT), reams of high-dimensional, streaming, and nonlinear data are being collected and curated to support decision-making. The detection of faults in these data is an important application in eMaintenance solutions, as it can facilitate maintenance decision-making. Early discovery of system faults may ensure the reliability and safety of industrial systems and reduce the risk of unplanned breakdowns. Complexities in the data, including high dimensionality, fast-flowing data streams, and high nonlinearity, impose stringent challenges on fault detection applications. From the data modelling perspective, high dimensionality may cause the notorious “curse of dimensionality” and lead to deterioration in the accuracy of fault detection algorithms. Fast-flowing data streams require algorithms to give real-time or near real-time responses upon the arrival of new samples. High nonlinearity requires fault detection approaches to have sufficiently expressive power and to avoid overfitting or underfitting problems. Most existing fault detection approaches work in relatively low-dimensional spaces. Theoretical studies on high-dimensional fault detection mainly focus on detecting anomalies on subspace projections. However, these models are either arbitrary in selecting subspaces or computationally intensive. To meet the requirements of fast-flowing data streams, several strategies have been proposed to adapt existing models to an online mode to make them applicable in stream data mining. But few studies have simultaneously tackled the challenges associated with high dimensionality and data streams. Existing nonlinear fault detection approaches cannot provide satisfactory performance in terms of smoothness, effectiveness, robustness and interpretability. New approaches are needed to address this issue. This research develops an Angle-based Subspace Anomaly Detection (ABSAD) approach to fault detection in high-dimensional data. The efficacy of the approach is demonstrated in analytical studies and numerical illustrations. Based on the sliding window strategy, the approach is extended to an online mode to detect faults in high-dimensional data streams. Experiments on synthetic datasets show the online extension can adapt to the time-varying behaviour of the monitored system and, hence, is applicable to dynamic fault detection. To deal with highly nonlinear data, the research proposes an Adaptive Kernel Density-based (Adaptive-KD) anomaly detection approach. Numerical illustrations show the approach’s superiority in terms of smoothness, effectiveness and robustness.

No description available.

Die stetige Entwicklung intelligenter Sensorsysteme erlaubt die Automatisierung und Verbesserung komplexer Prozess- und Geschäftsentscheidungen in vielfältigen Anwendungsszenarien. Sensoren können zum Beispiel zur Bestimmung optimaler Wartungstermine oder zur Steuerung von Produktionslinien genutzt werden. Ein grundlegendes Problem bereitet dabei die Sensordatenqualität, die durch Umwelteinflüsse und Sensorausfälle beschränkt wird. Ziel der vorliegenden Arbeit ist die Entwicklung eines Datenqualitätsmodells, das Anwendungen und Datenkonsumenten Qualitätsinformationen für eine umfassende Bewertung unsicherer Sensordaten zur Verfügung stellt. Neben Datenstrukturen zur effizienten Datenqualitätsverwaltung in Datenströmen und Datenbanken wird eine umfassende Datenqualitätsalgebra zur Berechnung der Qualität von Datenverarbeitungsergebnissen vorgestellt. Darüber hinaus werden Methoden zur Datenqualitätsverbesserung entwickelt, die speziell auf die Anforderungen der Sensordatenverarbeitung angepasst sind. Die Arbeit wird durch Ansätze zur nutzerfreundlichen Datenqualitätsanfrage und -visualisierung vervollständigt.

Sensors in smart-item environments capture data about product conditions and usage to support business decisions as well as production automation processes. A challenging issue in this application area is the restricted quality of sensor data due to limited sensor precision and sensor failures. Moreover, data stream processing to meet resource constraints in streaming environments introduces additional noise and decreases the data quality. In order to avoid wrong business decisions due to dirty data, quality characteristics have to be captured, processed, and provided to the respective business task. However, the issue of how to efficiently provide applications with information about data quality is still an open research problem. In this article, we address this problem by presenting a flexible model for the propagation and processing of data quality. The comprehensive analysis of common data stream processing operators and their impact on data quality allows a fruitful data evaluation and diminishes incorrect business decisions. Further, we propose the data quality model control to adapt the data quality granularity to the data stream interestingness.

This thesis deals with the data mining in data stream which represents fast developing area of information technology. The text describes common principles of data mining, explains what data stream is and shows methods for its preprocessing and algorithms for following data mining. The special attention is given to the VFDT and the CVDT algorithm. The next mentioned are the spatiotemporal data and related data mining. The second part describes the design and implementation of the application for classification over spatiotemporal data stream represented by road traffic data and following prediction of spatiotemporal events (traffic-jams). The classification is performed by the VFDT and CVFDT algorithm. The application has been tested on the data set obtained by the simulation tool SUMO.

Data mining (DM) is the process of finding patterns and relationships in databases. The breakthrough in computer technologies triggered a massive growth in data collected and maintained by organisations. In many applications, these data arrive continuously in large volumes as a sequence of instances known as a data stream. Mining these data is known as stream data mining. Due to the large amount of data arriving in a data stream, each record is normally expected to be processed only once. Moreover, this process can be carried out on different sites in the organisation simultaneously making the problem distributed in nature. Distributed stream data mining poses many challenges to the data mining community including scalability and coping with changes in the underlying concept over time. In this thesis, the author hypothesizes that learning classifier systems (LCSs) - a class of classification algorithms - have the potential to work efficiently in distributed stream data mining. LCSs are an incremental learner, and being evolutionary based they are inherently adaptive. However, they suffer from two main drawbacks that hinder their use as fast data mining algorithms. First, they require a large population size, which slows down the processing of arriving instances. Second, they require a large number of parameter settings, some of them are very sensitive to the nature of the learning problem. As a result, it becomes difficult to choose a right setup for totally unknown problems. The aim of this thesis is to attack these two problems in LCS, with a specific focus on UCS - a supervised evolutionary learning classifier system. UCS is chosen as it has been tested extensively on classification tasks and it is the supervised version of XCS, a state of the art LCS. In this thesis, the architectural design for a distributed stream data mining system will be first introduced. The problems that UCS should face in a distributed data stream task are confirmed through a large number of experiments with UCS and the proposed architectural design. To overcome the problem of large population sizes, the idea of using a Neural Network to represent the action in UCS is proposed. This new system - called NLCS { was validated experimentally using a small fixed population size and has shown a large reduction in the population size needed to learn the underlying concept in the data. An adaptive version of NLCS called ANCS is then introduced. The adaptive version dynamically controls the population size of NLCS. A comprehensive analysis of the behaviour of ANCS revealed interesting patterns in the behaviour of the parameters, which motivated an ensemble version of the algorithm with 9 nodes, each using a different parameter setting. In total they cover all patterns of behaviour noticed in the system. A voting gate is used for the ensemble. The resultant ensemble does not require any parameter setting, and showed better performance on all datasets tested. The thesis concludes with testing the ANCS system in the architectural design for distributed environments proposed earlier. The contributions of the thesis are: (1) reducing the UCS population size by an order of magnitude using a neural representation; (2) introducing a mechanism for adapting the population size; (3) proposing an ensemble method that does not require parameter setting; and primarily (4) showing that the proposed LCS can work efficiently for distributed stream data mining tasks.

Data mining (DM) is the process of finding patterns and relationships in databases. The breakthrough in computer technologies triggered a massive growth in data collected and maintained by organisations. In many applications, these data arrive continuously in large volumes as a sequence of instances known as a data stream. Mining these data is known as stream data mining. Due to the large amount of data arriving in a data stream, each record is normally expected to be processed only once. Moreover, this process can be carried out on different sites in the organisation simultaneously making the problem distributed in nature. Distributed stream data mining poses many challenges to the data mining community including scalability and coping with changes in the underlying concept over time. In this thesis, the author hypothesizes that learning classifier systems (LCSs) - a class of classification algorithms - have the potential to work efficiently in distributed stream data mining. LCSs are an incremental learner, and being evolutionary based they are inherently adaptive. However, they suffer from two main drawbacks that hinder their use as fast data mining algorithms. First, they require a large population size, which slows down the processing of arriving instances. Second, they require a large number of parameter settings, some of them are very sensitive to the nature of the learning problem. As a result, it becomes difficult to choose a right setup for totally unknown problems. The aim of this thesis is to attack these two problems in LCS, with a specific focus on UCS - a supervised evolutionary learning classifier system. UCS is chosen as it has been tested extensively on classification tasks and it is the supervised version of XCS, a state of the art LCS. In this thesis, the architectural design for a distributed stream data mining system will be first introduced. The problems that UCS should face in a distributed data stream task are confirmed through a large number of experiments with UCS and the proposed architectural design. To overcome the problem of large population sizes, the idea of using a Neural Network to represent the action in UCS is proposed. This new system - called NLCS { was validated experimentally using a small fixed population size and has shown a large reduction in the population size needed to learn the underlying concept in the data. An adaptive version of NLCS called ANCS is then introduced. The adaptive version dynamically controls the population size of NLCS. A comprehensive analysis of the behaviour of ANCS revealed interesting patterns in the behaviour of the parameters, which motivated an ensemble version of the algorithm with 9 nodes, each using a different parameter setting. In total they cover all patterns of behaviour noticed in the system. A voting gate is used for the ensemble. The resultant ensemble does not require any parameter setting, and showed better performance on all datasets tested. The thesis concludes with testing the ANCS system in the architectural design for distributed environments proposed earlier. The contributions of the thesis are: (1) reducing the UCS population size by an order of magnitude using a neural representation; (2) introducing a mechanism for adapting the population size; (3) proposing an ensemble method that does not require parameter setting; and primarily (4) showing that the proposed LCS can work efficiently for distributed stream data mining tasks.