Ontological Semantics Methods for Automatic Downgrading

Abstract

The purpose of the present work is to discuss the application of ontological semantics to the automatic downgrading of texts. The process of downgrading is particularly important to the government. This work shows how ontological semantics reduces the manual effort needed for downgrading texts. Ontological semantics is a recent approach that made use of the recent theories of computational semantics that appeared in the 1990s and were not made of use before. The reader concludes that ontological semantics succeeds in automatically processing and accordingly downgrading a large number of texts. It succeeds in processing most cases of ellipsis, word sense ambiguities, unrecorded words, metonymy, etc. Two modules of the system are used: first, the one that runs a partial analysis of a given text depending only on static knowledge resources, and second, the one that runs a full analysis of any problematic parts of the text by building text meaning representations (TMRs) of these parts. There are still some cases which are not resolved by NLP tools yet, most of which are cases of non-literal language, such as inferences, metaphors, etc.