May the Force Be with Your Copy Mechanism: Enhanced Supervised-Copy Method for Natural Language Generation
This addresses a specific bottleneck in copy mechanisms for natural language generation, offering an incremental improvement for tasks like summarization and data-to-text generation.
The paper tackled the problem of identifying which words to copy in neural sequence-to-sequence models with a copy mechanism, proposing a supervised approach that re-defines the objective function to use source sequences and target vocabularies as guidance. The result showed enhanced copying quality and improved abstractness in data-to-text generation and abstractive summarization tasks.
Recent neural sequence-to-sequence models with a copy mechanism have achieved remarkable progress in various text generation tasks. These models addressed out-of-vocabulary problems and facilitated the generation of rare words. However, the identification of the word which needs to be copied is difficult, as observed by prior copy models, which suffer from incorrect generation and lacking abstractness. In this paper, we propose a novel supervised approach of a copy network that helps the model decide which words need to be copied and which need to be generated. Specifically, we re-define the objective function, which leverages source sequences and target vocabularies as guidance for copying. The experimental results on data-to-text generation and abstractive summarization tasks verify that our approach enhances the copying quality and improves the degree of abstractness.