Graph-based Filtering of Out-of-Vocabulary Words for Encoder-Decoder Models
This addresses vocabulary selection for encoder-decoder models in NLP tasks like translation and error correction, but it is incremental as it builds on existing frequency-based methods.
The paper tackles the problem of selecting vocabulary words for encoder-decoder models by using co-occurrence information via the HITS algorithm, resulting in a 0.56-point BLEU score improvement for Japanese-to-English translation and a 1.48-point F0.5-measure gain for English grammatical error correction.
Encoder-decoder models typically only employ words that are frequently used in the training corpus to reduce the computational costs and exclude noise. However, this vocabulary set may still include words that interfere with learning in encoder-decoder models. This paper proposes a method for selecting more suitable words for learning encoders by utilizing not only frequency, but also co-occurrence information, which we capture using the HITS algorithm. We apply our proposed method to two tasks: machine translation and grammatical error correction. For Japanese-to-English translation, this method achieves a BLEU score that is 0.56 points more than that of a baseline. It also outperforms the baseline method for English grammatical error correction, with an F0.5-measure that is 1.48 points higher.