This volume constitutes a comprehensive self-contained course on source encoding. This is a rapidly developing field and the purpose of this book is to present the theory from its beginnings to the latest developments, some of which appear in book form for the first time. The major differences between this volume and previously published works is that here information retrieval is incorporated into source coding instead of discussing it separately. Second, this volume places an emphasis on the trade-off between complexity and the quality of coding; i.e. what is the price of achieving a maximum degree of data compression? Third, special attention is paid to universal families which contain a good compressing map for every source in a set. The volume presents a new algorithm for retrieval, which is optimal with respect to both program length and running time, and algorithms for hashing and adaptive on-line compressing. All the main tools of source coding and data compression such as Shannon, Ziv--Lempel, Gilbert--Moore codes, Kolmogorov complexity epsilon-entropy, lexicographic and digital search, are discussed. Moreover, data compression methods are described for developing short programs for partially specified Boolean functions, short formulas for threshold functions, identification keys, stochastic algorithms for finding the occurrence of a word in a text, and T-independent sets. For researchers and graduate students of information theory and theoretical computer science. The book will also serve as a useful reference for communication engineers and database designers.

Objectives Computer and communication practice relies on data compression and dictionary search methods. They lean on a rapidly developing theory. Its exposition from a new viewpoint is the purpose of the book. We start from the very beginning and finish with the latest achievements of the theory, some of them in print for the first time. The book is intended for serving as both a monograph and a self-contained textbook. Information retrieval is the subject of the treatises by D. Knuth (1973) and K. Mehlhorn (1987). Data compression is the subject of source coding. It is a chapter of information theory. Its up-to-date state is presented in the books of Storer (1988), Lynch (1985), T. Bell et al. (1990). The difference between them and the present book is as follows. First. We include information retrieval into source coding instead of discussing it separately. Information-theoretic methods proved to be very effective in information search. Second. For many years the target of the source coding theory was the estimation of the maximal degree of the data compression. This target is practically bit today. The sought degree is now known for most of the sources. We believe that the next target must be the estimation of the price of approaching that degree. So, we are concerned with trade-off between complexity and quality of coding. Third. We pay special attention to universal families that contain a good com pressing map for every source in a set."