George F. Foster

Confidence estimation for machine translation

We present a detailed study of confidence estimation for machine translation. Various methods for determining whether MT output is correct are investigated, for both whole sentences and words. Since the notion of correctness is not intuitively clear in this context, different ways of defining it are proposed. We present results on data from the NIST 2003 Chinese-to-English MT evaluation.

Mixture-Model Adaptation for SMT

We describe a mixture-model approach to adapting a Statistical Machine Translation System for new domains, using weights that depend on text distances to mixture components. We investigate a number of variants on this approach, including cross-domain versus dynamic adaptation; linear versus loglinear mixtures; language and translation model adaptation; different methods of assigning weights; and granularity of the source unit being adapted to. The best methods achieve gains of approximately one BLEU percentage point over a state-of-the art non-adapted baseline system.

Phrasetable Smoothing for Statistical Machine Translation

We discuss different strategies for smoothing the phrasetable in Statistical MT, and give results over a range of translation settings. We show that any type of smoothing is a better idea than the relative-frequency estimates that are often used. The best smoothing techniques yield consistent gains of approximately 1% (absolute) according to the BLEU metric.

Target-Text Mediated Interactive Machine Translation

The use of Machine Translation as a tool for professional or other highly skilled translators is for the most part currently limited to postediting arrangements in which the translator invokes MT when desired and then manually cleans up the results. A theoretically promising but hitherto largely unsuccessful alternative to postediting for this application is interactive machine translation (IMT), in which the translator and MT system work in tandem. We argue that past failures to make IMT viable as a tool for skilled translators have been the result of an infelicitous mode of interaction rather than any inherent flaw in the idea. As a solution, we propose a new style of IMT in which the target text under construction serves as the medium of communication between an MT system and its user. We describe the design, implementation, and performance of an automatic word completion system for translators which is intended to demonstrate the feasibility of the proposed approach, albeit in a very rudimentary form.

TransType: a computer-aided translation typing system

This paper describes the embedding of a statistical translation system within a text editor to produce TRANSTYPE, a system that watches over the user as he or she types a translation and repeatedly suggests completions for the text already entered. This innovative Embedded Machine Translation system is thus a specialized means of helping produce high quality translations.

Confidence estimation for translation prediction

The purpose of this work is to investigate the use of machine learning approaches for confidence estimation within a statistical machine translation application. Specifically, we attempt to learn probabilities of correctness for various model predictions, based on the native probabilites (i.e. the probabilites given by the original model) and on features of the current context. Our experiments were conducted using three original translation models and two types of neural nets (single-layer and multilayer perceptrons) for the confidence estimation task.

Stabilizing Minimum Error Rate Training

The most commonly used method for training feature weights in statistical machine translation (SMT) systems is Ochs minimum error rate training (MERT) procedure. A well-known problem with Ochs procedure is that it tends to be sensitive to small changes in the system, particularly when the number of features is large. In this paper, we quantify the stability of Ochs procedure by supplying different random seeds to a core component of the procedure (Powells algorithm). We show that for systems with many features, there is extensive variation in outcomes, both on the development data and on the test data. We analyze the causes of this variation and propose modifications to the MERT procedure that improve stability while helping performance on test data.

Unit Completion for a Computer-aided Translation Typing System

This work is in the context of, a system thatwatches over the users as they type a translation andrepeatedly suggests completions for the text already entered.The users may either accept, modify, or ignore these suggestions. Wedescribe the design, implementation, and performance of aprototype which suggests completions of units of texts that arelonger than one word.

PORTAGE: A Phrase-Based Machine Translation System

This paper describes the participation of the Portage team at NRC Canada in the shared task of ACL 2005 Workshop on Building and Using Parallel Texts. We discuss Portage, a statistical phrase-based machine translation system, and present experimental results on the four language pairs of the shared task. First, we focus on the French-English task using multiple resources and techniques. Then we describe our contribution on the Finnish-English, Spanish-English and German-English language pairs using the provided data for the shared task.

A maximum entropy/minimum divergence translation model

I present empirical comparisons between a linear combination of standard statistical language and translation models and an equivalent Maximum Entropy/Minimum Divergence (MEMD) model, using several different methods for automatic feature selection. The MEMD model significantly outperforms the standard model in test corpus perplexity, even though it has far fewer parameters.