DeepAI AI Chat
Log In Sign Up

Maximizing Stylistic Control and Semantic Accuracy in NLG: Personality Variation and Discourse Contrast

by   Vrindavan Harrison, et al.
University of California Santa Cruz

Neural generation methods for task-oriented dialogue typically generate from a meaning representation that is populated using a database of domain information, such as a table of data describing a restaurant. While earlier work focused solely on the semantic fidelity of outputs, recent work has started to explore methods for controlling the style of the generated text while simultaneously achieving semantic accuracy. Here we experiment with two stylistic benchmark tasks, generating language that exhibits variation in personality, and generating discourse contrast. We report a huge performance improvement in both stylistic control and semantic accuracy over the state of the art on both of these benchmarks. We test several different models and show that putting stylistic conditioning in the decoder and eliminating the semantic re-ranker used in earlier models results in more than 15 points higher BLEU for Personality, with a reduction of semantic error to near zero. We also report an improvement from .75 to .81 in controlling contrast and a reduction in semantic error from 16


page 1

page 2

page 3

page 4


Controlling Personality Style in Dialogue with Zero-Shot Prompt-Based Learning

Prompt-based or in-context learning has achieved high zero-shot performa...

Controlling Personality-Based Stylistic Variation with Neural Natural Language Generators

Natural language generators for task-oriented dialogue must effectively ...

Controlling Dialogue Generation with Semantic Exemplars

Dialogue systems pretrained with large language models generate locally ...

Controlling Linguistic Style Aspects in Neural Language Generation

Most work on neural natural language generation (NNLG) focus on controll...

Constrained Decoding for Neural NLG from Compositional Representations in Task-Oriented Dialogue

Generating fluent natural language responses from structured semantic re...

Multitask Learning for Class-Imbalanced Discourse Classification

Small class-imbalanced datasets, common in many high-level semantic task...

Tracking Discourse Influence in Darknet Forums

This technical report documents our efforts in addressing the tasks set ...