A Two-Stage Parsing Method for Text-Level Discourse Analysis

Abstract

Previous work introduced transition-based algorithms to form a unified architecture of parsing rhetorical structures (including span, nuclearity and relation), but did not achieve satisfactory performance. In this paper, we propose that transition-based model is more appropriate for parsing the naked discourse tree (i.e., identifying span and nuclearity) due to data sparsity. At the same time, we argue that relation labeling can benefit from naked tree structure and should be treated elaborately with consideration of three kinds of relations including within-sentence, across-sentence and across-paragraph relations. Thus, we design a pipelined two-stage parsing method for generating an RST tree from text. Experimental results show that our method achieves state-of-the-art performance, especially on span and nuclearity identification.

Predicting Discourse Structure using Distant Supervision from Sentiment

“…Consistent with previous work, e.g., Wang et al (2017); Joty et al (2015) and following the recent analysis by Morey et al (2017), our key metric is the average micro precision on span level, computed as the global overlap of the discourse structure prediction and the gold structure. We traverse both discourse trees tree pred i and tree gold i of each document i in post-order and compute:…”

“…Two-stage Parser: current state-of-the-art parser by Wang et al (2017). Employs two separate SVM classifiers for structure/nuclearity and relations, reaching the best performance for structure and nuclearity.…”

A Unified Linear-Time Framework for Sentence-Level Discourse Parsing

“…Another important distinction between existing methods is whether they employ a greedy transition-based algorithm (Marcu, 1999;Hirst, 2012, 2014b;Ji and Eisenstein, 2014;Braud et al, 2017;Li et al, 2016;Wang et al, 2017) or a globally optimized chart parsing algorithm (Soricut and Marcu, 2003;Li et al, 2014;Joty et al, 2015). Transition-based parsers build the tree incrementally by making a series of shiftreduce action decisions.…”

Hierarchical Pointer Net Parsing

“…Earlier methods have mostly utilized handcrafted lexical and syntactic features (Soricut and Marcu, 2003;Feng and Hirst, 2014;Joty et al, 2015;Wang et al, 2017). Recent approaches have shown competitive results with neural models that are able to automatically learn the feature representations in an end-to-end fashion (Ji and Eisenstein, 2014;Li et al, 2014).…”

“…Existing approaches to parsing can be distinguished based on whether they employ a greedy transition-based algorithm (Marcu, 1999;Zhang and Nivre, 2011;Wang et al, 2017) or a globally optimized algorithm such as graph-based methods for dependency parsing (Eisner, 1996) or chart parsing for discourse (Soricut and Marcu, 2003;Joty et al, 2015). Transition-based parsers build the tree incrementally by making a series * Linlin Liu is under the Joint PhD Program between Alibaba and Nanyang Technological University.…”