Enhancing Task-Oriented Dialogues With Chitchat: a Comparative Study Based on Lexical Diversity and Divergence
Published in ASRU, 2023
Comparative study based on entropy of recent Task-Oriented Dialogue datasets enhanced with chitchat.
Download here
Publications
A Unified Approach to Emotion Detection and Task-Oriented Dialogue Modeling
Published in Under Review, 2023
A unified approach to user emotion detection and TOD modeling. Tasks are learned with a single language modeling objective and lead to improved task success rates and more empathetic responses.
Chitchat as Interference: Adding User Backstories to Task-Oriented Dialogues
Published in Under Review, 2023
Novel chitchat augmentation for MultiWOZ using in-context learning with Llama-2-70B. A chatty user adds elements of backstory to their request, interfering with the smooth progression of the task.
Towards More Natural Dialogues: Integrating Open-Domain Dialogue Skills into Task-Oriented Agents.
Published in YRRSDS, workshop associated with SIGDial, 2023
Position paper on the intersection between chitchat and task-oriented dialogues (TODs), with a focus on integrating capabilities typically associated with chitchat systems into task-oriented agents.
Download here
Searching for Snippets of Open Domain Dialogue in Task-Oriented Dialogue Datasets
Published in Arxiv, 2022
Topic modeling analysis of Schema-Guided Dialogues and MultiWOZ, where we search for topics that are most similar to a set of social-talk-related keywords.
Download here
Question Answering in Natural Language: the Special Case of Temporal Expressions
Published in Student Research Workshop at RANLP, 2021
SQuAD-like approach to temporal Question Answering, with a curated datasets of 6000 examples and a LSTM-based model with an attention mechanism.
Download here