Tamás Gaál

WFSC: a new weighted finite state compiler

This article presents a new tool, WFSC, for creating, manipulating, and applying weighted finite state automata. It inherits some powerful features from Xeroxs non-weighted XFST tool and represents a continuation of Xeroxs work in the field of finite state automata over two decades. The design is generic: algorithms work on abstract components of automata and on a generic abstract semiring, and are independent of their concrete realizations. Applications can access WFSCs functions through an API or create automata through an end-user interface, either from an enumeration of their states and transitions or from rational expressions.

Is this Finite-State Transducer Sequentiable?

Sequentiality is a desirable property of finite state transducers: such transducers are optimal for time efficiency. Not all transducers are sequentiable. Sequentialization algorithms of finite state transducers do not recognize whether a transducer is sequentiable or not and simply do not ever halt when it is not. Choffrut proved that sequentiality of finite state transducers is decidable. B�al et al. have proposed squaring to decide sequentiality. We propose a different procedure, which, with �-closure extension, is able to handle letter transducers with arbitrary �-ambiguities, too. Our algorithm is more economical than squaring, in terms of size. In different cases of non-sequentiability necessary and sufficient conditions of the ambiguity class of the transducer can be observed. These ambiguities can be mapped bijectively to particular basic patterns in the structure of the transducer. These patterns can be recognized, using finite state methods, in any transducer.

Extended Sequentialization of Transducers

Sequential transducers, introduced by Schutzenberger [5], have advantageous computational properties. A sequential transducer is deterministic with respect to its input. Not all transducers can be sequentialized: but if one can be, it means time, and, often, space optimality. This article extends the subsequentialization algorithm of Mohri [3,4] for previously untreated classes of transducers. We — change the representation of final p-strings, — extend the sequentialization to input e labels and their closures, — handle the unknown symbol..

LABELLING MULTI-TAPE AUTOMATA WITH CONSTRAINED SYMBOL CLASSES

Rational relations are a powerful model used in many domains such as natural language processing. In this article, we propose a new model of finite state automata: multi-tape automata with symbol classes and identity or non-identity constraints. This model generalizes classical multi-tape automata, as well as automata and transducers with extended alphabet. We define this model in terms of a constraint satisfaction problem and discuss a problem occurring when handling the projection operation. Finally, we describe its implementation and results of a performance test.

Deciding sequentiability of finite-state transducers by finite-state pattern-matching

Sequentiality (input-side determinism) is a desirable property of finite-state transducers: such transducers are optimal for time efficiency. Not all transducers are sequentiable and those that are may not be sequential. Sequentialization algorithms of finite-state transducers do not recognize whether a transducer is sequentiable or not and simply do not ever halt when it is not. Choffrut proved that sequentiality of finite-state transducers is decidable. Beal et al. (in: D. Gonnet, G. Panario, A. Viola (Eds.), Proceedings of LATIN 2000, Lecture Notes in Computer Science, Vol. C1776, Springer, Heidelberg, 2000, p. 397) have proposed squaring to decide sequentiality. We propose a different procedure, which, with e-closure extension, is able to handle letter transducers with arbitrary e-ambiguities, too. Our algorithm is more economical than squaring, in terms of size. In different cases of non-sequentiability, necessary and sufficient conditions of one of the four possible ambiguity classes of the transducer can be observed. These ambiguities can be mapped bijectively to particular basic patterns in the structure of the transducer. The non-presence of both the infinitely ambiguous and the unboundedly ambiguous patterns is the condition of sequentiability. These patterns can be recognized, using finite-state methods, in any transducer. The method shows both sequentiability and, if present, sequentiality on the given side of the transducer.