# Learning to Plan and Realize Separately for Open-Ended Dialogue Systems

arXiv: https://arxiv.org/abs/2009.12506 · HTML: https://mlsanity.com/paper/2009.12506/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 26 Sep 2020.

## Built on by (most cited first, 6 of 6)

- [Towards Standard Criteria for human evaluation of Chatbots: A Survey](https://mlsanity.com/paper/2105.11197/) (2021)
- [Redefining Proactivity for Information Seeking Dialogue](https://mlsanity.com/paper/2410.15297/) (2024)
- [Semantic-Enhanced Explainable Finetuning for Open-Domain Dialogues](https://mlsanity.com/paper/2106.03065/) (2021)
- [Towards a Neural Era in Dialogue Management for Collaboration: A Literature Survey](https://mlsanity.com/paper/2307.09021/) (2023)
- [Use of a Taxonomy of Empathetic Response Intents to Control and Interpret Empathy in Neural Chatbots](https://mlsanity.com/paper/2305.10096/) (2023)
- [HERAKLES: Hierarchical Skill Compilation for Open-ended LLM Agents](https://mlsanity.com/paper/2508.14751/) (2025)

## Abstract (arXiv)

Achieving true human-like ability to conduct a conversation remains an elusive goal for open-ended dialogue systems. We posit this is because extant approaches towards natural language generation (NLG) are typically construed as end-to-end architectures that do not adequately model human generation processes. To investigate, we decouple generation into two separate phases: planning and realization. In the planning phase, we train two planners to generate plans for response utterances. The realization phase uses response plans to produce an appropriate response. Through rigorous evaluations, both automated and human, we demonstrate that decoupling the process into planning and realization performs better than an end-to-end approach.