Skip to main content

Research Repository

Advanced Search

Hierarchical reinforcement learning for situated natural language generation

Dethlefs, Nina; Cuayáhuitl, Heriberto

Authors

Heriberto Cuayáhuitl



Abstract

Natural Language Generation systems in interactive settings often face a multitude of choices, given that the communicative effect of each utterance they generate depends crucially on the interplay between its physical circumstances, addressee and interaction history. This is particularly true in interactive and situated settings. In this paper we present a novel approach for situated Natural Language Generation in dialogue that is based on hierarchical reinforcement learning and learns the best utterance for a context by optimisation through trial and error. The model is trained from human–human corpus data and learns particularly to balance the trade-off between efficiency and detail in giving instructions: the user needs to be given sufficient information to execute their task, but without exceeding their cognitive load. We present results from simulation and a task-based human evaluation study comparing two different versions of hierarchical reinforcement learning: One operates using a hierarchy of policies with a large state space and local knowledge, and the other additionally shares knowledge across generation subtasks to enhance performance. Results show that sharing knowledge across subtasks achieves better performance than learning in isolation, leading to smoother and more successful interactions that are better perceived by human users.

Citation

Dethlefs, N., & Cuayáhuitl, H. (2015). Hierarchical reinforcement learning for situated natural language generation. Natural language engineering, 21(3), 391-435. https://doi.org/10.1017/S1351324913000375

Acceptance Date Nov 18, 2013
Online Publication Date Jan 10, 2014
Publication Date 2015-05
Deposit Date Jan 19, 2016
Publicly Available Date Mar 29, 2024
Journal Natural language engineering
Print ISSN 1351-3249
Electronic ISSN 1469-8110
Publisher Cambridge University Press
Peer Reviewed Peer Reviewed
Volume 21
Issue 3
Pages 391-435
DOI https://doi.org/10.1017/S1351324913000375
Keywords Natural language generation
Public URL https://hull-repository.worktribe.com/output/384273
Publisher URL http://journals.cambridge.org/action/displayAbstract?fromPage=online&aid=9719879&fulltextType=RA&fileId=S1351324913000375
Additional Information Author's accepted manuscript of article published in: Natural language engineering, 2015, v.21, issue 3.

Files






You might also like



Downloadable Citations