Filling Knowledge Gaps in a Broad-Coverage Machine Translation System

TL;DR

This paper discusses methods to fill knowledge gaps in a broad-coverage Japanese-English machine translation system, enhancing its robustness and coverage by leveraging statistical techniques to improve translation quality across diverse domains.

Contribution

It introduces techniques for filling knowledge gaps in KBMT systems, demonstrating their effectiveness on a broad-coverage Japanese-English translation system.

Findings

Improved translation quality across multiple domains

Effective use of statistical techniques for knowledge gap filling

Enhanced system robustness in the absence of definitive knowledge

Abstract

Knowledge-based machine translation (KBMT) techniques yield high quality in domains with detailed semantic models, limited vocabulary, and controlled input grammar. Scaling up along these dimensions means acquiring large knowledge resources. It also means behaving reasonably when definitive knowledge is not yet available. This paper describes how we can fill various KBMT knowledge gaps, often using robust statistical techniques. We describe quantitative and qualitative results from JAPANGLOSS, a broad-coverage Japanese-English MT system.