Essentials of Pattern Mining

Knowledge Discovery from Databases (KDD) and Data Mining (DM) are general terms for a research area that deals with constructing models from data for predictive, descriptive, or summarizing purposes. Existing text books provide a broad overview of these problems and usually devote most attention to topics such as clustering and classification. Their discussions of pattern mining are often restricted to the basics of itemset mining and association rule discovery. However, pattern mining is more than frequent itemset mining. Pattern mining can be seen as the generalization of itemset mining to any task where a symbolic summary of data is needed. As a result of this, pattern mining is in fact at the core of many KDD problems but existing text books contrast those settings instead of relating them. Additionally, much progress in pattern mining has been made in recent years, in particular on how to select small sets of patterns and exploit patterns in other mining tasks, such as classification. This work is not yet organized in an easily accessible textbook. This book will organize many recent results in a pattern mining-centric way, hence providing a convenient, more advanced introduction to mining and using patterns. The book focusses on the "essentials", which is to say that we cover the full range - from quality and interestingness measures to constraints and their properties, to algorithms that perform pattern mining efficiently and those that select relevant subsets, and finally how to build pattern-based models - without going into too much depth on particular specialized solutions. We have divided the contents into a number of short, largely wellcontained chapters. At the end of each chapter, the reader will be able to implement basic pattern mining algorithms, or extend them towards more sophisticated solutions. This makes this book particularly well suited as an introduction for non-data miners, or as teaching material to computer science students. The field is too vast to do all developments justice in a single accessible book, yet we offer extensive lists of related reading material that both newcomers and experienced researchers can use to deepen their understanding.