Global-to-local Memory Pointer Networks for Task-Oriented Dialogue

By Chien-Sheng Wu, Richard Socher, Caiming Xiong. 2019

173 citations recorded by the source corpus.

End-to-end task-oriented dialogue is challenging since knowledge bases are usually large, dynamic and hard to incorporate into a learning framework. We propose the global-to-local memory pointer (GLMP) networks to address this issue. In our model, a global memory encoder and a local memory decoder are proposed to share external knowledge. The encoder encodes dialogue history, modifies global contextual representation, and generates a global memory pointer. The decoder first generates a sketch response with unfilled slots. Next, it passes the global memory pointer to filter the external knowledge for relevant information, then instantiates the slots via the local memory pointers. We empirically show that our model can improve copy accuracy and mitigate the common out-of-vocabulary problem. As a result, GLMP is able to improve over the previous state-of-the-art models in both simulated bAbI Dialogue dataset and human-human Stanford Multi-domain Dialogue dataset on automatic and human evaluation.

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

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].”
Survey of Hallucination in Natural Language Generation (2022, background): “Recently, there has emerged another line of research in task-oriented dialogue, which is to build a single end-to-end system rather than connecting several modules [87, 150].”
UnifiedSKG: Unifying and Multi-Tasking Structured Knowledge Grounding with Text-to-Text Language Models (2022, methodology): “Ruiqi For the KVRET dataset, instead of the version used in our main tables, we re-run another more widely used pre-processed version (Madotto et al., 2018; Wu et al., 2019; Qin et al., 2020) on T5-base, T5-large and T5-3b.”
Continual Learning for Task-oriented Dialogue System with Iterative Network Pruning, Expanding and Masking (2021, positive): “In this paper, we adopt the global-to-local memory pointer networks (GLMP) (Wu et al., 2019) as base model, which has shown powerful performance in TDS.”
Transferable Multi-Domain State Generator for Task-Oriented Dialogue Systems (2019, background): “, 2015), hardgated copy (Gulcehre et al., 2016; Madotto et al., 2018; Wu et al., 2019) and soft-gated copy (See et al.”
Neural Assistant: Joint Action Prediction, Response Generation, and Latent Knowledge Reasoning (2019, unknown): “[27] use a generative model instead of a text classification model but they along with previous work [1, 12, 5] work with much smaller knowledge bases where unlike in our case, full softmax attention over the knowledge base is computationally feasible.”
Contextualize Knowledge Bases with Transformer for End-to-end Task-Oriented Dialogue Systems (2020, unknown): “(Wu et al., 2019) proposes a global-tolocal memory pointer network to reduce the noise caused by KB.”
BiToD: A Bilingual Multi-Domain Dataset For Task-Oriented Dialogue Modeling (2021, background): “Recently, there has been growing interest in developing deep learning-based 19 end-to-end ToD systems [2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16] because they can handle 20 complex dialogue patterns with minimal hand-crafted rules.”
Amalgamating Knowledge from Two Teachers for Task-oriented Dialogue System with Adversarial Training (2020, methodology): “GLMP has achieved a strong improvement over both BLEU and entity F1 scores over the previous models, which is mainly beneﬁted from its global and local memory pointers to guide the KB attention and response generation.”
Slot-consistent NLG for Task-oriented Dialogue Systems with Iterative Rectification Network (2020, methodology): “A recent method in (Wu et al., 2019) deﬁnes an auxiliary loss that checks if the object words exist in the expected system response of a task-oriented dialogue system.”
Exploring Auxiliary Reasoning Tasks for Task-oriented Dialog Systems with Meta Cooperative Learning (2021, methodology): “First, the Seq2Seq model struggles to effectively reason over the complex KB and integrate the KB entities into dialog generation, making it unstable to generate the dialog responses (Wu, Socher, and Xiong 2019).”
A Survey of Knowledge-enhanced Text Generation (2020, methodology): “com/MaZhiyuanBUAA/Seq2BFforDialogueGeneration [130] PyTorch AAAI 2018 https://github.”
Knowledge-Enriched Transformer for Emotion Detection in Textual Conversations (2019, background): “…Pineau, 2018; Liu et al., 2018; Moghe et al., 2018; Dinan et al., 2019; Zhong et al., 2019), task-oriented dialogue systems (Madotto et al., 2018; Wu et al., 2019; He et al., 2019) and question answering systems (Kiddon et al., 2016; Hao et al., 2017; Sun et al., 2018; Mihaylov and Frank, 2018).”
Dynamic Fusion Network for Multi-Domain End-to-end Task-Oriented Dialog (2020, methodology): “Follow the prior work (Eric et al., 2017; Madotto et al., 2018; Wen et al., 2018; Wu et al., 2019a; Qin et al., 2019b), we adopt the BLEU and Micro Entity F1 metrics to evaluate model performance.”
Dual Dynamic Memory Network for End-to-End Multi-turn Task-oriented Dialog Systems (2020, methodology): “Compared with GLMP, DDMN is able to generate more proper and natural responses even in the last few turns during the conversation.”
Learning Knowledge Bases with Parameters for Task-Oriented Dialogue Systems (2020, background): “…(Eric et al., 2017a; Madotto et al., 2018; Qin et al., 2019, 2020; Banerjee and Khapra, 2019; Neelakantan et al., 2019) or an intermediate API call to retrieve part of the KB ( API+KB ) (Bordes and Weston, 2017; Eric and Manning, 2017; Madotto et al., 2018; Reddy et al., 2019; Wu et al., 2019b).”