license: mit
task_categories:
- text-generation
- summarization
- question-answering
- text-classification
- feature-extraction
- fill-mask
language:
- en
tags:
- debate
- argument
- argument mining
pretty_name: O
size_categories:
- 1M<n<10M
Dataset Card for Dataset Name
Dataset Description
- Homepage:
- Repository: TBA
- Paper: TBA
- Leaderboard:
- Point of Contact:
Dataset Summary
This dataset is a gargantoum follow-up to DebateSum, which includes a ton of improvements
Among those improvements are the following:
- Massively increased size (about 25X the size of DebateSum), including nearly all debate evidence ever open sourced over the past 20 years from High School and College Public Forum, Policy, and Lincoln Douglas debate leagues
- Far more metadata: Lots of now columns indicating everything from the number of times a piece of evidence has been seen (a good heuristic for evidence quality) to the teams and tournaments and rounds where a piece of evidence was deployed
- Better deduplication and parsing techniques, including better accounting of the hierarchical nature that debaters use for underlining evidence
Supported Tasks and Leaderboards
This dataset is useful for text generation, summarization, information retrieval, question answering, and related tasks. This dataset is further highly useful as a "trustworthy" dataset. All evidence within it has corresponding citations and is
Languages
English with very minor exceptions (i.e. evidence from performance cases using non-english evidence to make anti-colonalist arguments)
Dataset Creation
Gathered from the OpenCaseList project
Source Data
Debate Evidence from NDCA/NDT debate leagues from 2002-2022.
Considerations for Using the Data
Discussion of Biases
Competitive debate at the highest levels has increasingly rewarded teams who cite particular subfields of philosophy. A partial list of these highly represented topics is given below
- Postmodernism
- Poststructuralism
- Frankfurt School
- Critical Theory
- Critical Race Theory
- Queer Theory
- Feminism
These cannons are dominated by so called "left-wing" thinkers and have mostly marginalized so called "right-wing" thinkers within them with some notable exceptions. Note that despite a strong "left-wing" bias, large swaths of left-wing thought, such as anarchism, are relatively absent
Beyond this, most of the evidence was gathered with the argument being made first, and the evidence found after-the-fact to support it. This means that while the evidence is almost all "truthful", a lot of important information may be omitted
Other Known Limitations
There are cases of academic dishonesty within this dataset (i.e. evidence that had specific insertions made by a debater which werent in the original text). It's also possible that the source had changed inbetween when it was cited and retrieved.
Licensing Information
MIT