Publication
Ein Generator mit Anfragesystem für strukturierte Wörterbücher zur Unterstützung von Texterkennung und Textanalyse
Thomas Kieninger; Rainer Hoch
DFKI, DFKI Documents (D), Vol. 93-08, 1993.
Abstract
This document describes the implementation of a structured dictionary for document analysis allowing access to lexical data by two distinct methods. On one hand, there is a technique called three-fold hashing which automatically selects the best of three provided hash functions according to the quality of recognition results. For this kind of access a regular expression can be specified dealing with noisy input. In addition, the system provides a concept of (lexical) views for pruning search space, i.e., additional knowledge sources are used for fast accessing the dictionary. On the other hand, complex queries over the members of a view can be initiated based on a special filter concept. The realization of such views is a major topic of our dictionary system. Furthermore, we differentiate between two main components of the system: the dictionary generator and the dictionary controller (run time system).