Hedge Logic Programming for Information Extraction and Transformation.

Accession number;01A0851582
Title;Hedge Logic Programming for Information Extraction and Transformation.
Author;YAMAMOTO AKIHIRO(Hokudai Chishikimediaraboratori)   ITO KIMIHITO(Hokudai Chishikimediaraboratori)   ISHINO AKIRA(Kyudai Daigakuinshisutemujohogakukenkyuin)   
Journal Title;Proceedings of the Annual Conference of JSAI
Journal Code:X0580A
ISSN:
VOL.15th;NO.Vol.2;PAGE.3E2.03,1-4(2001)
Figure&Table&Reference;REF.18
Pub. Country;Japan
Language;Japanese
Abstract;In order to give foundations of extracting information from semi-structured documents and transforming them, we investigate a variant of logic programming in which a program treats hedges directly. Such logic programming is called hedge logic programming. A hedge, which sometimes called a forest, is a data structure obtained by parsing semi-structured documents. It is an intermediate of a string pattern and a first-order term. In this paper we formalize semi-structured documents and hedges so that logic programs can treat them directly as arguments of predicates. We give same conditions which are needed when hedge logic programming is applied for information extraction and transformation. For information extraction, we also consider generalization of hedges with giving an anti-untification algorithm for pairs of simple hedges. We clarify the relation among hedge logic programming, the original logic programming, and Elementary Formal Systems(EFS). (author abst.)