Information extraction from ancient hellenic texts using automatic knowledge acquisition from lexical definitions
Ioanna Malagardi, John Kontos · Neural, Parallel & Scientific Computations archive · 2001
The present paper presents the implementation of a system for information extraction from Ancient Hellenic Texts. The computer processing of phrases from the Gortys Law text aiming at the extraction and the semantic characterization of elements of legal rules from this text is used as an example. The computer implementation is based on the logic programming language Prolog. The system consists of an electronic lexicon, rules of syntax, and logical facts that codify prerequisite knowledge. The semantic relations that are usually implicit are discovered with the use of prerequisite knowledge. The problem of reconstructing the prerequisite knowledge of the writers and readers of texts of such an ancient period is very hard. It was solved by finding knowledge involved in analogous contemporary situations. This is accomplished by the translation of the Ancient Hellenic words into Modern Greek and using the machine-readable definition of them. The prerequisite knowledge is then extracted automatically from these definitions. The operation of the system is illustrated by presenting its performance in answering different kinds of natural language questions automatically. It is concluded that contemporary natural language processing methods can be applied fruitfully to Ancient Hellenic texts.