Key-Value Retrieval Networks for Task-Oriented Dialogue

By Mihail Eric, Lakshmi. Krishnan, Francois Charette, Christopher D. Manning. 2017

442 citations recorded by the source corpus.

Neural task-oriented dialogue systems often struggle to smoothly interface with a knowledge base. In this work, we seek to address this problem by proposing a new neural dialogue agent that is able to effectively sustain grounded, multi-domain discourse through a novel key-value retrieval mechanism. The model is end-to-end differentiable and does not need to explicitly model dialogue state or belief trackers. We also release a new dataset of 3,031 dialogues that are grounded through underlying knowledge bases and span three distinct tasks in the in-car personal assistant space: calendar scheduling, weather information retrieval, and point-of-interest navigation. Our architecture is simultaneously trained on data from all domains and significantly outperforms a competitive rule-based system and other existing neural dialogue architectures on the provided domains according to both automatic and human evaluation metrics.

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

Towards Exploiting Background Knowledge for Building Conversation Systems (2018, background): “…dialog datasets even before the advent of large-scale datasets for deep learning (Raux et al., 2005; Seneff et al., 1991) as well as in recent times (Rojas-Barahona et al., 2017; Williams et al., 2016; Eric et al., 2017) where datasets include small sized knowledge graphs as background knowledge.”
A Simple Language Model for Task-Oriented Dialogue (2020, methodology): “Several works used both dialogue history and knowledge bases as input and optimized neural encoder-decoder models to generate or retrieve system responses without modular supervision [13, 62, 29, 54, 56].”
UnifiedSKG: Unifying and Multi-Tasking Structured Knowledge Grounding with Text-to-Text Language Models (2022, background): “, 2019) 1024 32 4 KVRET (Eric et al., 2017) 1024 32 4 Conversational SParC (Yu et al.”
Neural Approaches to Conversational AI (2018, neutral): “ystem responses are ﬁrst collected and then used to train multiple components of a dialogue system in order to maximize prediction accuracy (Bordes et al., 2017; Wen et al., 2017; Yang et al., 2017b; Eric et al., 2017; Madotto et al., 2018; Wu et al., 2018). Wen et al. (2017) introduced a modular ne”
Lifelong and Continual Learning Dialogue Systems: Learning during Conversation (2021, background): “, (Chicago, CityOf, USA)] (Eric and Manning 2017; Young et al. 2018; Zhou et al. 2018b) to support information-seeking conversations, relevant response generation and help users with product recommendations.”
Incremental Learning from Scratch for Task-Oriented Dialogue Systems (2019, methodology): “Existing work on data-driven task-oriented dialogue systems includes generation based meth-ods (Wen et al., 2016; Eric and Manning, 2017) and retrieval based methods (Bordes et al., 2016; Williams et al., 2017; Li et al., 2017).”
MultiWOZ - A Large-Scale Multi-Domain Wizard-of-Oz Dataset for Task-Oriented Dialogue Modelling (2018, methodology): “, 2016), and even end-to-end dialogue modelling (Zhao and Eskenazi, 2016; Wen et al., 2017; Eric et al., 2017).”
Mem2Seq: Effectively Incorporating Knowledge Bases into End-to-End Task-Oriented Dialog Systems (2018, methodology): “Note that the original In-Car Assis- tant F1 scores reported in Eric et al. (2017) uses the entities in their canonicalized forms, which are not calculated based on real entity value.”
Global-to-local Memory Pointer Networks for Task-Oriented Dialogue (2019, methodology): “In the human-human dialogue dataset (Eric et al., 2017), GLMP is able to surpass the previous state of the art on both automatic and human evaluation, which further confirms the effectiveness of our double pointers usage.”
A Survey on Dialogue Systems: Recent Advances and New Frontiers (2017, background): “[15] augmented existing recurrent network architectures with a differentiable attention-based key-value retrieval mechanism over the entries of a knowledge base, which is inspired by key-value memory networks[44].”
Task-Oriented Dialog Systems that Consider Multiple Appropriate Responses under the Same Context (2019, background): “2017) to end-to-end trainable dialog models (Zhao et al. 2017; Wen et al. 2017; Eric and Manning 2017).”
Survey on evaluation methods for dialogue systems (2019, background): “1 Example dialogue where the driver can query the agenda via a voice command (Eric et al., 2017).”
Sequicity: Simplifying Task-oriented Dialogue Systems with Single Sequence-to-Sequence Architectures (2018, methodology): “U t R t to B t − 1 R t − 1 U t by compressing key points for the task completion given past dialogues; on the other hand, because bspans revisit context by only handling the B t with a ﬁxed length, the time complexity of TSCP is only O ( T ) , comparing O ( T 2 ) in (Eric and Manning, 2017a).”
RoMe: A Robust Metric for Evaluating Natural Language Generation (2022, methodology): “…from various domains of NLG such093 as knowledge graph based language generation094 dataset (KELM (Agarwal et al., 2021)), dialogue095 datasets (Eric et al., 2017; Chaudhuri et al., 2021),096 the NLG2017 challenge dataset (Shimorina et al.,097 2018), structured data to language generation098…”
BiToD: A Bilingual Multi-Domain Dataset For Task-Oriented Dialogue Modeling (2021, background): “Recently, there has been growing interest in developing deep learning-based end-to-end ToD systems [2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16] because they can handle complex dialogue patterns with minimal hand-crafted rules.”
Paraphrase Augmented Task-Oriented Dialog Generation (2020, background): “Task-oriented dialog systems that are applied to restaurant reservation and ticket booking have at-tracted extensive attention recently (Young et al., 2013; Wen et al., 2017; Bordes et al., 2016; Eric and Manning, 2017).”
TOD-BERT: Pre-trained Natural Language Understanding for Task-Oriented Dialogue (2020, background): “• SMD (Eric and Manning, 2017): Stanford multi-domain dialogue is an in-car personal assistant dataset, comprising 3,301 dialogues and three domains: calendar scheduling, weather information retrieval, and point-of-interest navigation.”
Lifelong and Interactive Learning of Factual Knowledge in Dialogues (2019, background): “Many task-oriented dialogue systems (Eric and Manning, 2017; Madotto et al., 2018) also use KBs to support information-seeking conversations.”
Lifelong Learning Dialogue Systems: Chatbots that Self-Learn On the Job (2020, background): “, (Chicago, CityOf, USA)] (Eric and Manning 2017; Madotto, Wu, and Fung 2018; Ghazvininejad et al. 2018; Le, Dymetman, and Renders 2016; Young et al. 2018; Long et al. 2017; Zhou et al. 2018b) to support information-seeking conversations and help users with product recommendations.”
Contextualize Knowledge Bases with Transformer for End-to-end Task-Oriented Dialogue Systems (2020, unknown): “On the SMD dataset, the BLEU score of COMET is higher than that of KB-Transformer by 3.4%.”