Tomasz Kowaltowski, Claudio L. Lucchesi, and Jorge Stol . Minimization of binary automata. In First South American String Processing Workshop, Belo Horizonte, Brasil, 1993.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....Finite state automata are used in various applications. One of the reasons for this is that they provide very compact representations of sets of strings. However, the size of an automaton measured in bytes can vary considerably depending on the storage method in use. Most of them are described in [5], a primary reference for all interested in automata compression. However, 5] does not provide sucient data on the in uence of particular methods on the size of the resulting automaton. We investigate that in this paper. We used only deterministic, acyclic automata in our experiments. However, ....

....this is that they provide very compact representations of sets of strings. However, the size of an automaton measured in bytes can vary considerably depending on the storage method in use. Most of them are described in [5] a primary reference for all interested in automata compression. However, [5] does not provide sucient data on the in uence of particular methods on the size of the resulting automaton. We investigate that in this paper. We used only deterministic, acyclic automata in our experiments. However, the methods we used do not depend on that feature. The automata we used were ....

[Article contains additional citation context not shown here]

Tomasz Kowaltowski, Claudio L. Lucchesi, and Jorge Stol . Minimization of binary automata. In First South American String Processing Workshop, Belo Horizonte, Brasil, 1993.

....from p to r, then there are 128 transitions leaving p. Using predicates, there are only two transitions leaving p (one labeled by a predicate which is only true of , and one labeled by : But note that similar space reductions can be achieved using failure transitions and related techniques (Kowaltowski, Lucchesi, and Stolfi, 1993; Kiraz, 1999; Daciuk, 2000) More interesting space reductions can be achieved in the case of transducers. The introduction of predicates with identity not only leads to transducers with fewer transitions, but also to transducers that have fewer states. This observation will be discussed in ....

Kowaltowski, Tomasz, Claudio L. Lucchesi, and Jorge Stolfi. 1993. Minimization of binary automata. In First South American String Processing Workshop, Belo Horizonte, Brasil.

....namely when input alphabet contains one symbol and in the case of acyclic DFA s (i.e. when the language accepted is finite) linear time algorithms are known. DFA PROBLEMS 11 In addition to being a fundamental problem, DFA minimization has practical applications in natural language processing [29], speech recognition and many others. DFA minimization has also been extensively studied on many parallel computation models, e.g. 22, 48, 23, 4] etc. J aJ a and Kosaraju [23] presented an efficient algorithm in the special case of one letter input alphabet on a meshconnected parallel computer. ....

....is in data compression. One way in which compression can be achieved by automata is to represent a finite set of sentences (say a subset of natural language sentences, or the words in a dictionary) by a DFA accepting the set. In this context, some nonstandard DFA models have been proposed, see [29]. In [29] an efficient heuristic algorithm has been proposed to minimize their model of DFA. It will be interesting to parallelize such algorithms since such algorithms are of practical importance. OPEN PROBLEMS We conclude with a short list of open problems that stand out in this survey. 1) ....

[Article contains additional citation context not shown here]

T. Kowaltowski, C. Lucchesi and J. Stolfi, "Minimization of binary automata ", Technical Report, Universidad Estadual de Campinas, SP. 1993.

....acceptors and transducers) play increasingly important role in natural language processing. Deterministic, acyclic, nite state automata are used mainly as representations of various kinds of dictionaries. Their main advantages are their small size as compared with the data they hold (see e.g. [3]) and the very fast lookup of strings in an automaton proportional to the length of the string. Deterministic, acyclic, nite state automata (DAFSA) are used in a variety of applications, including DNA sequencing, computer virus detection, and VLSA design. In natural language processing, they ....

Tomasz Kowaltowski, Claudio L. Lucchesi, and Jorge Stol. Minimization of binary automata. In First South American String Processing Workshop, Belo Horizonte, Brasil, 1993.

....The drawback of deterministic finite state automata as a formalisation of regular languages is that such DFA are in general not very compact. In fact, this is the reason that in many implementations of large DFA a somewhat less efficient but much more compact format is chosen (Daciuk, 1998; Kowaltowski, Lucchesi, and Stolfi, 1993). Many other formalisations of regular languages exist. For instance, regular expressions are another very popular device to present regular languages. Such regular expressions can be implemented fairly efficiently as well (for instance, the regular expression matching in Perl is done without ....

....of registers. This allows for a much more compact representation. Moreover, efficiency of implementation is hardly affected. However, the drawback is that there is little known about the construction of a minimal representative for 30 Gertjan van Noord such augmented finite state devices. In Kowaltowski, Lucchesi, and Stolfi (1993) binary automata are introduced as a representation for regular languages. Again, very compact and fairly efficient representations are possible. The authors discuss a few heuristics to minimize such binary automata. No general algorithms are known though. An important question therefore is ....

Kowaltowski, Tomasz, Cl audio L. Lucchesi, and Jorge Stolfi. 1993. Minimization of binary automata. Technical Report Relat orio Technico DCC-22/93, Departamento de Ciencia de Computacao, Universidade Estudual de Campinas.

....checkers and advisers, and mentioned other possible applications. Some of the important properties of finite automata which are explored in all of these applications are the compactness of the representation and the speed with which the information can be retrieved (see [5] In a companion paper ([4]) we study how to reduce even further the size of the representation without loss of access efficiency. Authors are listed in alphabetical order. In this work we focus on another application of such automata, namely the debugging of natural language vocabularies. Such vocabularies are large ....

Kowaltowski, T., Lucchesi, C. L., Stolfi, J., "Minimization of binary automata", in this issue.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC