OfekGlick commited on
Commit
29af869
1 Parent(s): fa40df6

Delete README.md

Browse files
Files changed (1) hide show
  1. README.md +0 -44
README.md DELETED
@@ -1,44 +0,0 @@
1
- DiscoEval Benchmark Datasets
2
- Overview
3
-
4
- The DiscoEval benchmark datasets consist of various tasks related to discourse analysis and natural language understanding. These datasets are designed to evaluate models' abilities in understanding and generating coherent discourse. The tasks include:
5
-
6
- Sentence Positioning
7
- Datasets: Arxiv, Wikipedia, Rocstory
8
- Description: In this task, the goal is to determine the correct positioning of a sentence within a given context or document.
9
-
10
- Binary Sentence Ordering
11
- Datasets: Arxiv, Wikipedia, Rocstory
12
- Description: Binary Sentence Ordering requires models to decide between two possible orders for a pair of sentences, determining which order provides a more coherent structure.
13
-
14
- Discourse Coherence
15
- Datasets: Ubuntu IRC channel, Wikipedia
16
- Description: This task assesses a model's ability to judge the coherence of a given discourse or conversation. Models must evaluate whether a sequence of sentences forms a coherent conversation or not.
17
-
18
- Sentence Section Prediction
19
- Dataset: Constructed from PeerRead
20
- Description: Sentence Section Prediction involves predicting the section or category to which a sentence belongs within a scientific paper, based on the content and context of the sentence.
21
-
22
- Discourse Relations
23
- Datasets: RST Discourse Treebank, Penn Discourse Treebank
24
- Description: This task focuses on identifying and classifying discourse relations between sentences or segments of text, helping to understand the structure and flow of discourse.
25
-
26
- Dataset Sources
27
-
28
- Arxiv: A collection of scientific papers and research articles.
29
- Wikipedia: An online encyclopedia containing a wide range of articles on various topics.
30
- Rocstory: A dataset consisting of fictional stories.
31
- Ubuntu IRC channel: Conversational data from the Ubuntu Internet Relay Chat (IRC) channel.
32
- PeerRead: Scientific paper dataset, often used for discourse-related tasks.
33
- RST Discourse Treebank: A collection of texts annotated with Rhetorical Structure Theory (RST) discourse relations.
34
- Penn Discourse Treebank: Another dataset with annotated discourse relations, helping to understand discourse structure.
35
-
36
- Use Cases
37
-
38
- The DiscoEval benchmark datasets can be used for training and evaluating various natural language understanding models, including those focused on discourse analysis, document structure prediction, and text generation.
39
- Citation
40
-
41
- Please refer to the respective sources and papers associated with each dataset for proper citation guidelines.
42
- License
43
-
44
- The datasets in the DiscoEval benchmark may have different licenses, so it is essential to review and comply with their individual licensing terms when using them in research or applications.