Scientific papers also follow a global structure, authors of the research or scientific paper have a scientific objective , create methods to reach the goal ,do comparative study of prior works on the topic and have various solutions substantiating the scientific goal. Similar to argumentation structure, scientific paper tries to make a knowledge claim.

Structure of Scientific Papers

Argumentation structure provides a way to arrange globally related sentences in a discourse. For example in persuasive essays people use claims and supporting premises.

The computational analysis of people’s arguments is called argument mining.

Algorithms for discovering argumentation structure often include classifiers for identifying claims, premises, or non-argumentation, together with relation classifiers for deciding if two spans have the SUPPORT, ATTACK, or neither relation.


Argumentation Structure

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


San Jose State University

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

Discourse Structure Parsing

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

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

Learn Before

Related

Learn After