Context Tracking Network: Graph-based Context Modeling for Implicit Discourse Relation Recognition

Zhang, Yingxue; Meng, Fandong; Li, Peng; Jian, Ping; Zhou, Jie

doi:10.18653/v1/2021.naacl-main.126

Cited by 13 publications

(10 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Some other hybrid neural networks have been designed through combining various models such as simplified topic model, gated convolutional network, factored tensor network, graph convolutional network, directed graphic model, sequence to sequence model, etc. [122,155,166,168]. For example, Xu et al [155] proposed a Topic Tensor Network (TTN) model which combines a simplified topic model, a gated convolutional network and a factored tensor network.…”

Section: Hybrid Neural Network Modelsmentioning

confidence: 99%

See 1 more Smart Citation

A Survey of Implicit Discourse Relation Recognition

Wang¹,

Wang²

2022

Preprint

View full text Add to dashboard Cite

A discourse containing one or more sentences describes daily issues and events for people to communicate their thoughts and opinions. As sentences are normally consist of multiple text segments, correct understanding of the theme of a discourse should take into consideration of the relations in between text segments. Although sometimes a connective exists in raw texts for conveying relations, it is more often the cases that no connective exists in between two text segments but some implicit relation does exist in between them. The task of implicit discourse relation recognition (IDRR) is to detect implicit relation and classify its sense between two text segments without a connective. Indeed, the IDRR task is important to diverse downstream natural language processing tasks, such as text summarization, machine translation and so on. This article provides a comprehensive and up-to-date survey for the IDRR task. We first summarize the task definition and data sources widely used in the field. We categorize the main solution approaches for the IDRR task from the viewpoint of its development history. In each solution category, we present and analyze the most representative methods, including their origins, ideas, strengths and weaknesses. We also present performance comparisons for those solutions experimented on a public corpus with standard data processing procedures. Finally, we discuss future research directions for discourse relation analysis.

show abstract

Section: Hybrid Neural Network Modelsmentioning

confidence: 99%

“…Zhang et al [168] proposed a Semantic Graph Convolutional Network (SGCN) to enhance the inter-argument semantic interaction. They first encoded each argument into a representation vector by a BiLSTM network.…”

Section: Hybrid Neural Network Modelsmentioning

confidence: 99%

A Survey of Implicit Discourse Relation Recognition

Wang¹,

Wang²

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…More recent work on implicit relation classification has adopted a graph-based context tracking network to model the necessary context for interpreting the discourse and has gained better performance on PDTB-2 (Zhang et al, 2021). In addition, the increase in the number of implicit relation instances in PDTB-3 (Prasad et al, 2019) has sparked more interest in exploring their recognition, such as Kim et al (2020) and Liang et al (2020).…”

Section: Previous Workmentioning

confidence: 99%

“…Kido and Aizawa 2016) and implicit (e.g. Liu et al 2016;Wang and Lan 2016;Rutherford et al 2017;Kim et al 2020;Liang et al 2020;Zhang et al 2021) discourse relation classification in English PDTB-2 (Prasad et al, 2008) and PDTB-3 (Prasad et al, 2019) as well as on the PDTB-style Chinese newswire corpus (CDTB, Zhou and Xue 2012;Zhou et al 2014) such as Schenk et al (2016) and Weiss and Bajec (2016).…”

Section: Introductionmentioning

confidence: 99%

DisCoDisCo at the DISRPT2021 Shared Task: A System for Discourse Segmentation, Classification, and Connective Detection

Gessler¹,

Behzad²,

Liu³

et al. 2021

Proceedings of the 2nd Shared Task on Discourse Relation Parsing and Treebanking (DISRPT 2021)

View full text Add to dashboard Cite

This paper describes our submission to the DISRPT2021 Shared Task on Discourse Unit Segmentation, Connective Detection, and Relation Classification. Our system, called Dis-CoDisCo, is a Transformer-based neural classifier which enhances contextualized word embeddings (CWEs) with hand-crafted features, relying on tokenwise sequence tagging for discourse segmentation and connective detection, and a feature-rich, encoder-less sentence pair classifier for relation classification. Our results for the first two tasks outperform SOTA scores from the previous 2019 shared task, and results on relation classification suggest strong performance on the new 2021 benchmark. Ablation tests show that including features beyond CWEs are helpful for both tasks, and a partial evaluation of multiple pre-trained Transformer-based language models indicates that models pre-trained on the Next Sentence Prediction (NSP) task are optimal for relation classification.

show abstract

“…The main difference between KG-augmented models is how they use messaging (Schlichtkrull et al, 2018; or edge/path aggregation (Lin et al, 2019;Bosselut and Choi, 2019;Ma et al, 2019;Wang et al, 2020a) used in NLP. Some works use GNN to model text structure (Yasunaga and Liang, 2020;Zhang et al, 2021), extract relations and entities (Xu et al, 2021a;, solve multi-hop question answering (Saxena et al, 2020;Fang et al, 2020), and so on. Our work is based on the Graph Attention Networks (GATs) (Veličković et al, 2018) framework.…”

Section: Commonsense Question Answeringmentioning

confidence: 99%

JointLK: Joint Reasoning with Language Models and Knowledge Graphs for Commonsense Question Answering

Sun¹,

Shi²,

Qi³

2021

Preprint

View full text Add to dashboard Cite

Existing KG-augmented models for question answering primarily focus on designing elaborate Graph Neural Networks (GNNs) to model knowledge graphs (KGs). However, they ignore (i) the effectively fusing and reasoning over question context representations and the KG representations, and (ii) automatically selecting relevant nodes from the noisy KGs during reasoning. In this paper, we propose a novel model, JointLK, which solves the above limitations through the joint reasoning of LMs and GNNs and the dynamic KGs pruning mechanism. Specifically, JointLK performs joint reasoning between the LMs and the GNNs through a novel dense bidirectional attention module, in which each question token attends on KG nodes and each KG node attends on question tokens, and the two modal representations fuse and update mutually by multi-step interactions. Then, the dynamic pruning module uses the attention weights generated by joint reasoning to recursively prune irrelevant KG nodes. Our results on the CommonsenseQA and OpenBookQA datasets demonstrate that our modal fusion and knowledge pruning methods can make better use of relevant knowledge for reasoning.

show abstract

Context Tracking Network: Graph-based Context Modeling for Implicit Discourse Relation Recognition

Cited by 13 publications

References 20 publications

A Survey of Implicit Discourse Relation Recognition

A Survey of Implicit Discourse Relation Recognition

DisCoDisCo at the DISRPT2021 Shared Task: A System for Discourse Segmentation, Classification, and Connective Detection

JointLK: Joint Reasoning with Language Models and Knowledge Graphs for Commonsense Question Answering

Contact Info

Product

Resources

About