Discourse Parsing

A machine reading comprehension (MRC) dataset with discourse structure built over multiparty dialog. Molweni's source samples from the Ubuntu Chat Corpus, including 10,000 dialogs comprising 88,303 utterances. Source: Molweni: A Challenge Multiparty Dialogues-based Machine Reading Comprehension Dataset with Discourse Structure

The Rhetorical Structure Theory (RST) Discourse Treebank consists of 385 Wall Street Journal articles from the Penn Treebank annotated with discourse structure in the RST framework along with human-generated extracts and abstracts associated with the source documents. In the RST framework (Mann and Thompson, 1988), a text's discourse structure can be represented as a tree in four aspects: (1) the leaves correspond to text fragments called elementary discourse units (the mininal discourse units); (2) the internal nodes of the tree correspond to contiguous text spans; (3) each node is characterized by its nuclearity, or essential unit of information; and (4) each node is also characterized by a rhetorical relation between two or more non-overlapping, adjacent text spans. Data The data in this release is divided into a training set (347 documents) and a test set (38 documents). All annotations were produced using a discourse annotation tool that can be downloaded from http://www.isi.edu/~marcu/discourse.

GUM is an open source multilayer English corpus of richly annotated texts from twelve text types. Annotations include: * Multiple POS tags, morphological features and lemmatization * Sentence segmentation and rough speech act * Document structure in TEI XML (paragraphs, headings, figures, etc.) * ISO date/time annotations * Speaker and addressee information (where relevant) * Constituent and dependency syntax * Information status (given, accessible, new, split antecedent) * Entity and coreference annotation, including bridging anaphora * Entity linking (Wikification) * Discourse parses in Rhetorical Structure Theory and discourse dependencies

AMALGUM is a machine annotated multilayer corpus following the same design and annotation layers as GUM, but substantially larger (around 4M tokens). The goal of this corpus is to close the gap between high quality, richly annotated, but small datasets, and the larger but shallowly annotated corpora that are often scraped from the Web.

This discourse treebank includes annotated instructional texts originally assembled at the Information Technology Research Institute, University of Brighton. This dataset contains 176 documents with an average of 32.6 EDUs for a total of 5744 EDUs and 53,250 words.

The DISRPT 2019 workshop introduces the first iteration of a cross-formalism shared task on discourse unit segmentation. Since all major discourse parsing frameworks imply a segmentation of texts into segments, learning segmentations for and from diverse resources is a promising area for converging methods and insights. We provide training, development and test datasets from all available languages and treebanks in the RST, SDRT and PDTB formalisms, using a uniform format. Because different corpora, languages and frameworks use different guidelines for segmentation, the shared task is meant to promote design of flexible methods for dealing with various guidelines, and help to push forward the discussion of standards for discourse units. For datasets which have treebanks, we will evaluate in two different scenarios: with and without gold syntax, or otherwise using provided automatic parses for comparison.

The DISRPT 2021 shared task, co-located with CODI 2021 at EMNLP, introduces the second iteration of a cross-formalism shared task on discourse unit segmentation and connective detection, as well as the first iteration of a cross-formalism discourse relation classification task. We provide training, development and test datasets from all available languages and treebanks in the RST, SDRT and PDTB formalisms, using a uniform format. Because different corpora, languages and frameworks use different guidelines, the shared task is meant to promote design of flexible methods for dealing with various guidelines, and help to push forward the discussion of standards for computational approaches to discourse relations. We include data for evaluation with and without gold syntax, or otherwise using provided automatic parses for comparison to gold syntax data.

The SPOT dataset contains 197 reviews originating from the Yelp'13 and IMDB collections ([1][2]), annotated with segment-level polarity labels (positive/neutral/negative). Annotations have been gathered on 2 levels of granulatiry: - Sentences - Elementary Discourse Units (EDUs), i.e. sub-sentence clauses produced by a state-of-the-art RST parser This dataset is intended to aid sentiment analysis research and, in particular, the evaluation of methods that attempt to predict sentiment on a fine-grained, segment-level basis. Source: SPOT

The Potsdam Commentary Corpus (PCC) is a corpus of 220 German newspaper commentaries (2.900 sentences, 44.000 tokens) taken from the online issues of the Märkische Allgemeine Zeitung (MAZ subcorpus) and Tagesspiegel (ProCon subcorpus) and is annotated with a range of different types of linguistic information. The central subcorpus that we are making publicly available consists of 176 MAZ texts, which are annotated with * Sentence Syntax * Coreference * Discourse Structure (RST & PDTB) * Aboutness topics

Parallel version of annotations in GUM RST v9.1.

RST corpus for Russian. * 233 documents * 2 genres (news & blogs) + relations annotation for academical texts * 28k elementary discourse units * 24 rhetorical relations