nlp:task_oriented_dialog
Table of Contents
Task Oriented Dialog
Papers
- Su et al 2021 - Multi-Task Pre-Training for Plug-and-Play Task-Oriented Dialogue System Introduces PPTOD model
- Discovering Latent Structure
Few-Shot Task Oriented Dialog
- Schema-Guided Dialog (SGD):2019 - Towards Scalable Multi-domain Conversational Agents: The Schema-Guided Dialogue Dataset
Zero Shot Task Oriented Dialog
Prompting and In-Context Learning
See also Response Generation, below.
Semi-supervised Methods
Domain Adaptation
Dialog State Tracking
“Dialog state tracking (DST) extracts users’ goals in task-oriented dialog systems, where dialog states are often represented in terms of a set of slot-value pairs” (from Dai 2021).
- MultiWoz
- MultiWOZ Also a data to text generation dataset.
- STAR: Ye et al 2021 - Slot Self-Attentive Dialogue State Tracking SOTA as of 2022
- Ye et al 2021 - MultiWOZ 2.4: A Multi-Domain Task-Oriented Dialogue Dataset with Essential Annotation Corrections to Improve State Tracking Evaluation Contains a comparison of SOTA methods on MultiWoz 2.1 and 2.4.
Response Generation
- With LLMs
Related Pages
nlp/task_oriented_dialog.txt · Last modified: 2023/09/08 18:39 by jmflanig