Task Completion Platform: A self-serve multi-domain goal oriented dialogue platform

Published 1 January 2016Open access

Paul Crook, Alex Marin, Vipul Agarwal, Khushboo Aggarwal, Tasos Anastasakos, Ravi Bikkula

Citations21

TL;DR

A multi-domain dialogue platform that can host and execute large numbers of goal-orientated dialogue tasks, and features a task configuration language, TaskForm, that allows the definition of each individual task to be decoupled from the overarching dialogue policy used by the platform to complete those tasks.

Abstract

Paul Crook, Alex Marin, Vipul Agarwal, Khushboo Aggarwal, Tasos Anastasakos, Ravi Bikkula, Daniel Boies, Asli Celikyilmaz, Senthilkumar Chandramohan, Zhaleh Feizollahi, Roman Holenstein, Minwoo Jeong, Omar Khan, Young-Bum Kim, Elizabeth Krawczyk, Xiaohu Liu, Danko Panic, Vasiliy Radostev, Nikhil Ramesh, Jean-Phillipe Robichaud, Alexandre Rochette, Logan Stromberg, Ruhi Sarikaya. Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Demonstrations. 2016.

Keywords

Computer Science

Computer Speech & LanguageThe RavenClaw dialog management framework: Architecture and systems

260 Citations2008Dan Bohus, Alexander I. Rudnicky

RavenClaw isolates the domain-specific aspects of the dialog control logic from domain-independent conversational skills, and in the process facilitates rapid development of mixed-initiative systems operating in complex, task-oriented domains.

Easy contextual intent prediction and slot detection

89 Citations2013Aditya Bhargava, Aslı Çelikyılmaz +2 more

This paper investigates the incorporation of context into the SLU tasks of intent prediction and slot detection, using a corpus that contains session-level information and finds that including features indicating the slots appearing in the previous utterances gives no significant increase in performance.

Deep contextual language understanding in spoken dialogue systems

27 Citations2015Chunxi Liu, Puyang Xu +1 more

This work shows with extensive experiments that the proposed joint learning framework generates state-of-the-art results for both classification and tagging, and the contextual modeling based on recurrent and external features significantly improves the context sensitivity of SLU models.

Resolving Referring Expressions in Conversational Dialogs for Natural User Interfaces

24 Citations2014Aslı Çelikyılmaz, Zhaleh Feizollahi +2 more

This work poses the problem as a classification task to correctly identify intended on-screen item(s) from user utterances with up to 90% accuracy using syntactic, semantic as well as context features from the display screen, and shows that the proposed model is robust to domain and screen layout changes.

View PDF

Hypotheses ranking for robust domain classification and tracking in dialogue systems

24 Citations2014Jean-Philippe Robichaud, Paul Crook +3 more

This work presents a novel application of hypothesis ranking (HR) for the task of domain detection in a multi-domain, multiturn dialog system and shows that HR can be trained to be robust to changes in the SLU.

Learning Domain-Independent Dialogue Policies via Ontology Parameterisation

21 Citations2015Zhuoran Wang, Tsung-Hsien Wen +2 more

The experimental results show that the policy optimised in a restaurant search domain using the proposed domain-independent representations can be deployed to a laptop sale domain, achieving a task success rate very close to that of the policy Optimised on in-domain dialogues.

View PDF

Multi-language hypotheses ranking and domain tracking for open domain dialogue systems

7 Citations2015Paul Crook, Jean-Philippe Robichaud +1 more

As the set of input features used by HR models are largely language independent a single, universal HR model can be used in place of language specific HR models with only a small loss in accuracy and generalise well to new unseen languages, especially related languages.

A universal model for flexible item selection in conversational dialogs

2 Citations2015Aslı Çelikyılmaz, Zhaleh Feizollahi +2 more

A single universal FIS model can be used in place of language specific FIS models with no loss in accuracy and it is shown that such a model can generalize well to new unseen languages with minimal loss in Accuracy.