Using Description Logics for Recognising Textual Entailment

TL;DR

This paper explores using Description Logics to address Recognising Textual Entailment by proposing a semantic representation and novel inference tasks, demonstrating a logical approach to natural language understanding.

Contribution

It introduces a new DL-based framework for RTE, including the definition of A-Box saturation and subgraph detection as key inference tasks.

Findings

Proposes a DL-based semantic representation for natural language.

Defines two novel inference tasks: A-Box saturation and subgraph detection.

Shows potential of DLs in improving RTE performance.

Abstract

The aim of this paper is to show how we can handle the Recognising Textual Entailment (RTE) task by using Description Logics (DLs). To do this, we propose a representation of natural language semantics in DLs inspired by existing representations in first-order logic. But our most significant contribution is the definition of two novel inference tasks: A-Box saturation and subgraph detection which are crucial for our approach to RTE.