Formal methods for mining structured objects (Q1946770)

A systematic approach to knowledge extraction from massive sets of data began in the 90s of the past century. This was the beginning of a new discipline, called ``knowledge discovery''; data mining techniques were developed to provide efficient automatizations of data analysis. This book deals with graph representations of data coming from sets of sequences, trees, or more general combinatorial structures. Concept graphs are used in knowledge discovery to draw information from such data sets. Formal concept analysis is devoted to the construction of a lattice of concepts by introducing a suitable Galois connection. Several examples and applications are given in the book, along with empirical experiments, with a view of providing new insights into the mining of various sorts of acyclic structured objects. The first chapter gives introductory material on the analysis of sequences and various mining techniques. Chapter 2 deals with concept lattices and Horn logic. Concept lattices are further discussed in Chapter 3 in connection with sequences, closure operators, Galois connections and the minimum support condition. Horn axiomatizations for sequences are the main topic of Chapter 4: an example is given on the reconstruction of the rules of a cellular automaton. Chapters 5 and 6 respectively focus on transformations on injective partial orders and general ones. Generalizations to other structured data are considered in Chapter 7. Chapter 8 provides a summary, followed by a comprehensive bibliography featuring over 140 entries. The book is a valuable addition to the literature on knowledge discovery.