RoSA '02 (Rouen, June 6-7, 2002)


Generalized Pattern Matching Statistics: Dynamical sources context


Jérémie BOURDON
bourdon@info.unicaen.fr
http://users.info.unicaen.fr/~bourdon
GREYC, Université de Caen, Bd Maréchal Juin, 14032 Caen Cedex, FRANCE
Joint work with B. Vallée


In pattern matching algorithms, a characteristic parameter is the number of occurrences of a given pattern in a random text of length n generated by a source. We consider here a generalization of the pattern matching problem in two ways. First, we deal with a generalized notion of pattern that encompasses classical patterns as well as "hidden patterns". Second, we consider a quite general probabilistic model of sources that may possess a high degree of correlations. Such sources are built with dynamical systems and are called dynamical sources. We determine the mean and the variance of the number of occurrences in this generalized pattern matching problem, and establish a property of concentration of distribution. These results are obtained via combinatorics, formal language techniques, and methods of analytic combinatorics based on generating operators and generating functions. The generating operators come from the dynamical system framework and generate themselves generating functions. The motivation to study this problem comes from an attempt at finding a reliable threshold for intrusion detections, from textual data processing applications, and from molecular biology.


Retour à la page d'accueil
Back to home page