Discourse Embellishment Using a Deep Encoder-Decoder Network
This addresses the problem of generating more literary text from lightweight NLG systems for computational storytellers, though it appears incremental as it builds on existing encoder-decoder methods.
The paper tackles the problem of automatically increasing lexical and syntactic complexity in text while preserving meaning, proposing textual embellishment as a new NLG task for computational storytelling. It presents promising first results using LSTM encoder-decoder networks trained on the WikiLarge dataset.
We suggest a new NLG task in the context of the discourse generation pipeline of computational storytelling systems. This task, textual embellishment, is defined by taking a text as input and generating a semantically equivalent output with increased lexical and syntactic complexity. Ideally, this would allow the authors of computational storytellers to implement just lightweight NLG systems and use a domain-independent embellishment module to translate its output into more literary text. We present promising first results on this task using LSTM Encoder-Decoder networks trained on the WikiLarge dataset. Furthermore, we introduce "Compiled Computer Tales", a corpus of computationally generated stories, that can be used to test the capabilities of embellishment algorithms.