Abstract:The discourse analysis task, which focuses on understanding the semantics of long text spans, has received increasing attention in recent years. As a critical component of discourse analysis, discourse relation recognition aims to identify the rhetorical relations between adjacent discourse units (e.g., clauses, sentences, and sentence groups), called arguments, in a document. Previous works focused on capturing the semantic interactions between arguments to recognize their discourse relations, ignoring important textual information in the surrounding contexts. However, in many cases, more than capturing semantic interactions from the texts of the two arguments are needed to identify their rhetorical relations, requiring mining more contextual clues. In this paper, we propose a method to convert the RST-style discourse trees in the training set into dependency-based trees and train a contextual evidence selector on these transformed structures. In this way, the selector can learn the ability to automatically pick critical textual information from the context (i.e., as evidence) for arguments to assist in discriminating their relations. Then we encode the arguments concatenated with corresponding evidence to obtain the enhanced argument representations. Finally, we combine original and enhanced argument representations to recognize their relations. In addition, we introduce auxiliary tasks to guide the training of the evidence selector to strengthen its selection ability. The experimental results on the Chinese CDTB dataset show that our method outperforms several state-of-the-art baselines in both micro and macro F1 scores.

Exploiting Discourse Relations for Sentiment Analysis.

Sentiment Analysis on the People’s Daily

Exploiting Hierarchical Discourse Structure for Review Sentiment Analysis

Unsupervised Discovery of Discourse Relations for Eliminating Intra-sentence Polarity Ambiguities.

Improve Discourse Dependency Parsing with Contextualized Representations

Exploiting the Unique Expression for Improved Sentiment Analysis in Software Engineering Text

Research of Using Structural Feature in Sentiment Analysis for Social Media Messages

Better Document-level Sentiment Analysis from RST Discourse Parsing

Mining The Relation Between Sentiment Expression And Target Using Dependency Of Words

Imbalanced Sentiment Classification Enhanced with Discourse Marker

Sentimental Direction Analysis: A Framework for Chinese Sentiment Computation and Resource Construction

Chinese Discourse Relation analysis and data annotation

Exploring Discourse Structures for Argument Impact Classification

A Study of the Application of Weight Distributing Method Combining Sentiment Dictionary and TF-IDF for Text Sentiment Analysis

The Effects of Discourse Connectives Prediction on Implicit Discourse Relation Recognition

Incorporating Contextual Evidence to Improve Implicit Discourse Relation Recognition in Chinese

The Chinese Discourse Parser

Exploring the use of word relation features for sentiment classification

Discourse Relations Detection Via a Mixed Generative-Discriminative Framework

A Novel Sentiment Polarity Detection Framework for Chinese