A typical syntactic pattern recognition (PR) problem involves comparing a noisy string with every element of a dictionary, H. The problem of classification can be greatly simplified if the dictionary is partitioned into a set of sub-dictionaries. In this case, the classification can be hierarchical -- the noisy string is first compared to a representative element of each sub-dictionary and the closest match within the sub-dictionary is subsequently located. Indeed, the entire problem of sub-dividing a set of strings into subsets where each subset contains "similar" strings has been referred to as the "String Taxonomy Problem". To our knowledge there is no reported solution to this problem (see footnote on Page 2). In this paper we shall present a learningautomaton based solution to string taxonomy. The solution utilizes the Object Migrating Automaton (OMA) whose power in clustering objects and images [33,35] has been reported. The power of the scheme for string taxonomy has been demons...

this paper is to present further results about the frequency and types of unknown words found in reallife corpora. We hope that the results of our study will be of some use in the development of NLP systems capable of dealing with realistic input

: In this paper an error-correcting parsing algorithm and its application to a postprocessing task in the context of automatic check processing is described. The proposed method has shown very good results in terms of recognition accuracy and execution speed on both real and synthetic data. 1 Introduction The recognition of machine printed characters has been intensively studied during the past years and significant progress has been made [1]. For example, there exist commercial OCR systems that achieve a correct recognition rate of over 99% today [2]. But depending on the particular application, such a high recognition rate may be still insufficient. In order to further improve recognition accuracy, contextual postprocessing is often very useful. Different contextual postprocessing methods have been proposed in the literature. They are based, for example, on n-gram statistics [3,4], or dictionary search [5,6]. A recent survey on contextual processing has been given in [7]. For earli...

In this paper we present a foundational basis for optimal and information theoretic syntactic pattern recognition. We do this by developing a rigorous model, M * , for channels which permit arbitrarily distributed substitution, deletion and insertion syntactic errors. More explicitly, if A is any finite alphabet and A * the set of words over A, we specify a stochastically consistent scheme by which a string U A * can be transformed into any Y A * by means of arbitrarily distributed substitution, deletion and insertion operations. The scheme is shown to be Functionally Complete and stochastically consistent. Apart from the synthesis aspects, we also deal with the analysis of such a model and derive a technique by which Pr[Y|U], the probability of receiving Y given that U was transmitted, can be computed in cubic time using dynamic programming. Experimental results which involve dictionaries with strings of lengths between 7 and 14 with an overall average noise of 39.75 % demons...

this paper is devoted to a summary of the state of the art in the domain of printed ocr (similar to the presentations in [Imp 91, Gov 90, Nad 84, Man 86]), by focussing attention essentially on the new orientations of ocr in the document recognition area. 2 Document Analysis Aspects

and J. Matheson, eds., Readings on the Principles and Applications

Comprendre un block de text 6;rit a la main c'est I'en mettre en correspondence avec une representation s~mantique. Nous dkrivons une approche pour lire un block de texte 6;rit Ii la main quand it y a certaines constraintes vagues dans la disposition typographique et Ie syntax du texte. Un sysreme pour lire des addresses postales 6;rit a la main est decrit com me une instance. Mots cles: Reconnaissance de caracteres, reconnaissance de l'6;riture Ii la main, system des classifieufS. A method for mapping a block of handwritten text into a symbolic representation is described. It is assumed that certain loose constraints are placed on the spatial layout and syntax of the text Early recognition of primitives guides the location of syntactic components. The instance of reading handwritten postal addresses is described, where the symbolic representation is ~ digit string (ZIP Code). Keywords:

Multi-level post-processing for Korean character recognition using