Learning End-to-End Goal-Oriented Dialog

By Antoine Bordes, Jason Aaron Edward Weston. 2016

788 citations recorded by the source corpus.

Traditional dialog systems used in goal-oriented applications require a lot of domain-specific handcrafting, which hinders scaling up to new domains. End-to-end dialog systems, in which all components are trained from the dialogs themselves, escape this limitation. But the encouraging success recently obtained in chit-chat dialog may not carry over to goal-oriented settings. This paper proposes a testbed to break down the strengths and shortcomings of end-to-end dialog systems in goal-oriented applications. Set in the context of restaurant reservation, our tasks require manipulating sentences and symbols, so as to properly conduct conversations, issue API calls and use the outputs of such calls. We show that an end-to-end dialog system based on Memory Networks can reach promising, yet imperfect, performance and learn to perform non-trivial operations. We confirm those results by comparing our system to a hand-crafted slot-filling baseline on data from the second Dialog State Tracking Challenge (Henderson et al., 2014a). We show similar result patterns on data extracted from an online concierge service.

Subfields: Artificial Intelligence, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Natural Language Processing

Citations in the corpus

Experience Grounds Language (2020, background): “one could imagine socializing with in anything more than transactional (Bordes et al., 2016) or extremely limited game scenarios (Lewis et al.”
Open-Domain Conversational Agents: Current Progress, Open Problems, and Future Directions (2020, background): “ Goal-oriented dialogue systems often have clear evaluation methodologies, e.g. task completion can be measured if the correct actions are taken (Hastie, 2012; Henderson, Thomson, and Williams, 2014; Bordes, Boureau, and Weston, 2017; El Asri et al., 2017; Wen et al., 2017). Chitchat tasks, such as ”
ACUTE-EVAL: Improved Dialogue Evaluation with Optimized Questions and Multi-turn Comparisons (2019, background): “Goal-oriented tasks typically have a clearer evaluation, e.g. task completion can be measured if the correct actions are taken (Hastie 2012; Henderson, Thomson, and Williams 2014; Bordes, Boureau, and Weston 2017; El Asri et al. 2017; Wen et al. 2017).”
Wizard of Wikipedia: Knowledge-Powered Conversational agents (2018, background): “ed-forward network. In the area of goal-directed dialogue, separate from open domain chit-chat, such as airline (El Asri et al., 2017) or restaurant booking (Henderson et al., 2014; Wen et al., 2016; Bordes et al., 2017), knowledge conditioning is typically employed by allowing access to a database ”
Towards Exploiting Background Knowledge for Building Conversation Systems (2018, background): “Some of these datasets contain transcripts of human-bot conversations (Williams et al., 2013; Henderson et al., 2014a,b) while others are created using a fixed set of natural language patterns (Bordes and Weston, 2017; Dodge et al., 2016).”
Learning Through Dialogue Interactions (2016, background): “Our work represents a first step in developing such end-to-end learned interactive dialogue agents.”
A Simple Language Model for Task-Oriented Dialogue (2020, background): “Some systems are mostly end-to-end, but still need to call out additional APIs or skip intermediate tasks like dialogue state tracking [5].”
Visual Dialog (2016, background): “Memory Network.”
Personalizing Dialogue Agents: I have a dog, do you have pets too? (2018, background): “Instead, the research community has focused on task-oriented communication, such as airline or restaurant booking (Bordes and Weston, 2016), or else single-turn informa-”
ParlAI: A Dialog Research Software Platform (2017, background): “, 2015), bAbI Dialog tasks (Bordes and Weston, 2016), Ubuntu (Lowe et al.”
Neural Approaches to Conversational AI (2018, neutral): “o training of the components is done separately in a supervised manner. This challenge is addressed by Dhingra et al. (2017) who proposed “soft” knowledge-base lookups; see Sec. 3.5 for more details. Bordes et al. (2017) treated dialogue system learning as the problem of learning a mapping from dial”
What makes a good conversation? How controllable attributes affect human judgments (2019, background): “Dialogue Dialogue evaluation is relatively well understood in goal-oriented tasks, where automated approaches can be coded by measuring task completion (Bordes et al., 2017; El Asri et al., 2017; Hastie, 2012; Henderson et al., 2014; Wen et al., 2017).”
Training Millions of Personalized Dialogue Agents (2018, background): “…dialogue systems, based on neural architectures like bidirectional LSTMs or Memory Networks (Sukhbaatar et al., 2015) trained directly by gradient descent on dialogue logs, have been showing promising performance in multiple contexts (Wen et al., 2016; Serban et al., 2016; Bordes et al., 2016).”
Generative Deep Neural Networks for Dialogue: A Short Review (2016, methodology): “In another line of work, researchers have started proposing alternative training and response selection criteria [Weston, 2016].”
Continual Learning for Natural Language Generation in Task-oriented Dialog Systems (2020, background): “As an essential part of task-oriented dialog systems (Wen et al., 2015b; Bordes et al., 2016), the task of Natural Language Generation (NLG) is to produce a natural language utterance containing the desired information given a semantic representation (so-called dialog act).”
Incremental Learning from Scratch for Task-Oriented Dialogue Systems (2019, methodology): “• IR: the basic tf-idf match model used in (Bordes et al., 2016; Dodge et al., 2015).”
End-to-End Offline Goal-Oriented Dialog Policy Learning via Policy Gradient (2017, methodology): “Recently, several supervised learning algorithms have been proposed to learn goal-oriented dialog policies [3, 28].”
MultiWOZ - A Large-Scale Multi-Domain Wizard-of-Oz Dataset for Task-Oriented Dialogue Modelling (2018, background): “These problems can either be addressed separately (Young et al., 2013), or jointly in an end-to-end fashion (Bordes et al., 2017; Wen et al., 2017; Li et al., 2017).”
Continuously Learning Neural Dialogue Management (2016, background): “Recently an emerging line of research has focused on training a network-based dialogue model, mostly in text-input schemes (Vinyals and Le, 2015; Serban et al., 2015; Wen et al., 2016; Bordes and Weston, 2016).”