Reliable Knowledge Discovery

Reliable Knowledge Discovery focuses on theory, methods, and techniques for RKDD, a new sub-field of KDD. It studies the theory and methods to assure the reliability and trustworthiness of discovered knowledge and to maintain the stability and consistency of knowledge discovery processes. RKDD has a broad spectrum of applications, especially in critical domains like medicine, finance, and military. Reliable Knowledge Discovery also presents methods and techniques for designing robust knowledge-discovery processes. Approaches to assessing the reliability of the discovered knowledge are introduced. Particular attention is paid to methods for reliable feature selection, reliable graph discovery, reliable classification, and stream mining. Estimating the data trustworthiness is covered in this volume as well. Case studies are provided in many chapters. Reliable Knowledge Discovery is designed for researchers and advanced-level students focused on computer science and electrical engineering as a secondary text or reference. Professionals working in this related field and KDD application developers will also find this book useful.

This comprehensive textbook on data mining details the unique steps of the knowledge discovery process that prescribes the sequence in which data mining projects should be performed, from problem and data understanding through data preprocessing to deployment of the results. This knowledge discovery approach is what distinguishes Data Mining from other texts in this area. The book provides a suite of exercises and includes links to instructional presentations. Furthermore, it contains appendices of relevant mathematical material.

The growth in the amount of data collected and generated has exploded in recent times with the widespread automation of various day-to-day activities, advances in high-level scienti?c and engineering research and the development of e?cient data collection tools. This has given rise to the need for automa- callyanalyzingthedatainordertoextractknowledgefromit,therebymaking the data potentially more useful. Knowledge discovery and data mining (KDD) is the process of identifying valid, novel, potentially useful and ultimately understandable patterns from massive data repositories. It is a multi-disciplinary topic, drawing from s- eral ?elds including expert systems, machine learning, intelligent databases, knowledge acquisition, case-based reasoning, pattern recognition and stat- tics. Many data mining systems have typically evolved around well-organized database systems (e.g., relational databases) containing relevant information. But, more and more, one ?nds relevant information hidden in unstructured text and in other complex forms. Mining in the domains of the world-wide web, bioinformatics, geoscienti?c data, and spatial and temporal applications comprise some illustrative examples in this regard. Discovery of knowledge, or potentially useful patterns, from such complex data often requires the - plication of advanced techniques that are better able to exploit the nature and representation of the data. Such advanced methods include, among o- ers, graph-based and tree-based approaches to relational learning, sequence mining, link-based classi?cation, Bayesian networks, hidden Markov models, neural networks, kernel-based methods, evolutionary algorithms, rough sets and fuzzy logic, and hybrid systems. Many of these methods are developed in the following chapters.

Knowledge Discovery and Measures of Interest is a reference book for knowledge discovery researchers, practitioners, and students. The knowledge discovery researcher will find that the material provides a theoretical foundation for measures of interest in data mining applications where diversity measures are used to rank summaries generated from databases. The knowledge discovery practitioner will find solid empirical evidence on which to base decisions regarding the choice of measures in data mining applications. The knowledge discovery student in a senior undergraduate or graduate course in databases and data mining will find the book is a good introduction to the concepts and techniques of measures of interest. In Knowledge Discovery and Measures of Interest, we study two closely related steps in any knowledge discovery system: the generation of discovered knowledge; and the interpretation and evaluation of discovered knowledge. In the generation step, we study data summarization, where a single dataset can be generalized in many different ways and to many different levels of granularity according to domain generalization graphs. In the interpretation and evaluation step, we study diversity measures as heuristics for ranking the interestingness of the summaries generated. The objective of this work is to introduce and evaluate a technique for ranking the interestingness of discovered patterns in data. It consists of four primary goals: To introduce domain generalization graphs for describing and guiding the generation of summaries from databases. To introduce and evaluate serial and parallel algorithms that traverse the domain generalization space described by the domain generalization graphs. To introduce and evaluate diversity measures as heuristic measures of interestingness for ranking summaries generated from databases. To develop the preliminary foundation for a theory of interestingness within the context of ranking summaries generated from databases. Knowledge Discovery and Measures of Interest is suitable as a secondary text in a graduate level course and as a reference for researchers and practitioners in industry.

Modern computer-based control systems are able to collect a large amount of information, display it to operators and store it in databases but the interpretation of the data and the subsequent decision making relies mainly on operators with little computer support. This book introduces developments in automatic analysis and interpretation of process-operational data both in real-time and over the operational history, and describes new concepts and methodologies for developing intelligent, state space-based systems for process monitoring, control and diagnosis. The book brings together new methods and algorithms from process monitoring and control, data mining and knowledge discovery, artificial intelligence, pattern recognition, and causal relationship discovery, as well as signal processing. It also provides a framework for integrating plant operators and supervisors into the design of process monitoring and control systems.

The importance of having ef cient and effective methods for data mining and kn- ledge discovery (DM&KD), to which the present book is devoted, grows every day and numerous such methods have been developed in recent decades. There exists a great variety of different settings for the main problem studied by data mining and knowledge discovery, and it seems that a very popular one is formulated in terms of binary attributes. In this setting, states of nature of the application area under consideration are described by Boolean vectors de ned on some attributes. That is, by data points de ned in the Boolean space of the attributes. It is postulated that there exists a partition of this space into two classes, which should be inferred as patterns on the attributes when only several data points are known, the so-called positive and negative training examples. The main problem in DM&KD is de ned as nding rules for recognizing (cl- sifying) new data points of unknown class, i. e. , deciding which of them are positive and which are negative. In other words, to infer the binary value of one more attribute, called the goal or class attribute. To solve this problem, some methods have been suggested which construct a Boolean function separating the two given sets of positive and negative training data points.

This book presents a specific and unified approach to Knowledge Discovery and Data Mining, termed IFN for Information Fuzzy Network methodology. Data Mining (DM) is the science of modelling and generalizing common patterns from large sets of multi-type data. DM is a part of KDD, which is the overall process for Knowledge Discovery in Databases. The accessibility and abundance of information today makes this a topic of particular importance and need. The book has three main parts complemented by appendices as well as software and project data that are accessible from the book's web site (http://www.eng.tau.ac.iV-maimonlifn-kdg£). Part I (Chapters 1-4) starts with the topic of KDD and DM in general and makes reference to other works in the field, especially those related to the information theoretic approach. The remainder of the book presents our work, starting with the IFN theory and algorithms. Part II (Chapters 5-6) discusses the methodology of application and includes case studies. Then in Part III (Chapters 7-9) a comparative study is presented, concluding with some advanced methods and open problems. The IFN, being a generic methodology, applies to a variety of fields, such as manufacturing, finance, health care, medicine, insurance, and human resources. The appendices expand on the relevant theoretical background and present descriptions of sample projects (including detailed results).