Epsilon-Lee/DRR-benchmark

Benchmarking discourse relation recognition (DRR)

typora-root-url
./

DRR-benchmark

[TOC]

Benchmarking discourse relation classification (DRR). This is a benchmarking test ground for discourse relation classification task, which is based on PDTB dataset.

Dataset

As for discourse-level analysis, two datasets are mainly used:

• PDTB: Penn Discourse Tree Bank;
• RST-DT: Rhetorical Structure Theory Discourse Tree Bank.

PDTB Split

• Train: 2-20 sections.
• Dev: 0-1 sections.
• Test: 21-22 sections.

Recent papers

a. Discourse relation classification/recognition

2017

• Adversarial Connective-exploiting Networks for Implicit Discourse Relation Classification, ACL 2017. Lianhui Qin et al. model

• (RNNAtt17) A Recurrent Neural Model with Attention for the Recognition of Chinese Implicit Discourse Relations, ACL 2017. model

  

• Improving Implicit Relation Recognition with Discourse-specific Word Embeddings, ACL 2017.

• A Systematic Study of Neural Discourse Models for Implicit Discourse Relation, EACL 2017.

• Do We Need Cross Validation for Discourse Relation Classification, EACL 2017.

• Discourse Relations and Conjoined VPs: Automated Sense Recognition, EACL 2017.

• (MTA17) Multi-task Attention-based Neural Networks for Implicit Discourse Relationship Representation and Identification, EMNLP 2017. Baidu Inc. model

  

2016

• (GRN16) Implicit Discourse Relation Detection via a Deep Architecture with Gated Relevance Network, ACL 2016. Fudan Univ. model

  

• Modelling the Interpretation of Discourse Connectives by Bayesian Pragmatics, ACL 2016.

• (LVRNN16) A Latent Variable Recurrent Neural Network for Discourse-Driven Language Models, NAACL 2016. model

  

• The Instantiation Discourse Relation: A Corpus Analysis of Its Properties and Improved Detection, NAACL 2016.

• Learning Connective-based Word Representations for Implicit Discourse Relation Identification, EMNLP 2016. model

• (VarDRR16) Variational Neural Discourse Relation Recognizer, EMNLP 2016. model

  

• (RR16) Recognizing Implicit Discourse Relations via Repeated Reading: Neural Networks with Multi-Level Attention, EMNLP 2016. model

  

• (SGN16) A Stacking Gated Neural Architecture for Implicit Discourse Relation Classification, EMNLP 2016. model

  

• Bilingually-constrained Synthetic Data for Implicit Discourse Relation Recognition, EMNLP 2016.

• (CCE16) Implicit Discourse Relation Recognition with Context-aware Character-enhanced Embeddings, COLING 2016. model

  

b. Discourse relation related tasks/problem formulations

• Neural Net Models of Open-domain Discourse Coherence, EMNLP 2017. Jiwei Li.