[ Skip to the content ]

Institute of Formal and Applied Linguistics Wiki

[ Back to the navigation ]

Table of Contents

Encouraging Consistent Translation Choices

Ferhan Ture, Douglas W. Oard, and Philip Resnik
NAACL 2012

Outline -- discussion

The list of discussed topics follows the outline of the paper:

Differences from Carpuat 2009

Human translators and one sense per discourse

Sec. 3. Exploratory analysis


Forced decoding

The choice and filtering of “cases”

Sec. 4. Approach

The actual experiments begin only now; the used data is different.

Choice of features

Feature weights

Meaning of the individual features

Requirements of the new features

Sec. 5. Evaluation and Discussion

Choice of baseline

Tuning the feature weights

The usage of two variants of the BLEU evaluation metric


Selection of the sentences for the analysis

BLEU deficiency

Sec. 6. Conclusions

Structural variation moderation

Choice of discourse context

Our conclusion

Nice paper with a very good idea that probably can improve translations, but with several arguments that are not backed up by sufficient evidence or clearly misleading. The initial analysis is done in a very precise and detailed way. The actual translation experiments show that adding new features helps, but lack some obvious steps, such as significance checking or actually proving that BLEU is wrong by using METEOR or similar metric.

Martin's remarks

[ Back to the navigation ] [ Back to the content ]