1. Find the discourse connectives (disambiguating them from non-discourse uses)
2. Find the two spans for each connective
3. Label the relationship between these spans
4. Assign a relation between every adjacent pair of sentences

Subtasks of PDTB discourse parsing

It's the task of detecting PDTB coherence relations between spans. Sometimes it's also called shallow discourse parsing.

University of Michigan - Ann Arbor

Given a sequence of sentences, automatically determine the coherence relations between them.

Discourse Structure Parsing

The Penn Discourse TreeBank (PDTB) is a second commonly used dataset that embodies model of coherence relations.

 Penn Discourse TreeBank (PDTB)

An on-going but a helpful book resource about NLP
https://web.stanford.edu/~jurafsky/slp3/

Speech and Language Processing (3rd ed. draft) 

It's the most commonly used model of discourse organization. In RST relations are defined between nucleus two spans of text, generally a nucleus and a satellite.

Rhetorical Structure theory (RST)

It corresponds to text spans of a sentence, clause or EDU phrase. It's the basic unit of RST.

Elementary Discourse Units (EDU)

EDU segmentation extracts the start and end of each EDU. The output should be segmentation of sentences. Modern systems generally use neural sequence models supervised by the gold EDU segmentation in datasets like the RST Discourse Treebank. A classic method can be to predict whether there is a break for every token.

EDU Segmentation

PDTB discourse parsing

Sentences in a discourse can be related to each other globally making the existence of global coherence in a discourse.

Examples are Narrative stories existing in Russian folktales, persuasive essays, scientific papers.

There are two kinds of global coherence structure that are studied computationally

- Argumentation structure

- The structure of scientific discourse


Learn Before

Related

Learn After