Finite-State Techniques
Automata, Transducers and Bimachines
Cambridge Tracts in Theoretical Computer Science Series

Finite-state methods are the most efficient mechanisms for analysing textual and symbolic data, providing elegant solutions for an immense number of practical problems in computational linguistics and computer science. This book for graduate students and researchers gives a complete coverage of the field, starting from a conceptual introduction and building to advanced topics and applications. The central finite-state technologies are introduced with mathematical rigour, ranging from simple finite-state automata to transducers and bimachines as 'input-output' devices. Special attention is given to the rich possibilities of simplifying, transforming and combining finite-state devices. All algorithms presented are accompanied by full correctness proofs and executable source code in a new programming language, C(M), which focuses on transparency of steps and simplicity of code. Thus, by enabling readers to obtain a deep formal understanding of the subject and to put finite-state methods to real use, this book closes the gap between theory and practice.

Part I. Formal Background: 1. Formal preliminaries; 2. Monoidal finite-state automata; 3. Classical finite-state automata and regular languages; 4. Monoidal multi-tape automata and finite-state transducers; 5. Deterministic transducers; 6. Bimachines; Part II. From Theory to Practice: 7. The C(M) language; 8. C(M) implementation of finite-state devices; 9. The Aho–Corasick algorithm; 10. The minimal deterministic finite-state automaton for a finite language; 11. Constructing finite-state devices for text rewriting; Bibliography; Index.

Stoyan Mihov is Associate Professor at the Bulgarian Academy of Sciences (IICT) and a lecturer at Sofia University. He has published several efficient automata constructions and approximate search methods, which are widely used for natural language processing and information retrieval. Dr Mihov has led the development of multiple award-winning systems for language and speech processing.
Klaus U. Schulz is Professor of Information and Language Processing at the Ludwig-Maximilians-Universität Munchen. He has published over 100 articles in distinct fields of computer science, with contributions in approximate search and transducer technology. He was head of many projects in text-correction and digital humanities, on both a national and European level.