Just-in-Time Grammar

In this paper, we attempt to explain why rule-based syntactic analysis systems have not so far worked as well as they could. This will motivate our approach, which is based on a narrowing of the focus of the rules, resulting in more accurate, less ambiguous, more efficient parsing and relative ease of acquisition. These rules are word-based, rather than construction-based, and are used to produce “Just in Time” grammars for specific inputs. We combine this grammar with an error detection and repair module that is enabled primarily by the dependency-directed capabilities of the basic parsing engine. Our approach is most closely related to the XTAG philosophy (Schabes, et. al., 1988). We discuss the similarities and differences between XTAG and our work while further affirming the benefits of lexicalized grammars..

• Type: Conference
• Year: 2003
• Where: MLMTA
• Authors: Stephen Beale, Sergei Nirenburg, Marjorie McShane
• Keywords: Basic Parsing Engine, Efficient Parsing, MLMTA 2003, MLMTA 2007, Rule-based Syntactic Analysis