A two-step neural text generator with sentence-level keyphrase planning and style control improves coherence and style fidelity over standard sequence-to-sequence baselines.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Sentence-Level Content Planning and Style Specification for Neural Text Generation
A two-step neural text generator with sentence-level keyphrase planning and style control improves coherence and style fidelity over standard sequence-to-sequence baselines.