Chandra Bhagavatula

3 paper cards in Mova's library · 2019–2020

Chandra Bhagavatula wrote 3 of the papers Mova reads, in 2019–2020; often with Ronan Le Bras, Yejin Choi, Keisuke Sakaguchi.

The profile is the library's match of the name; two people with one name are not always told apart.

Ideas Mova took (1)

From the ideas ledger: an idea tried, kept or queued, or measured with no gain, and what it gave.

kept — Partial-input controls for every course, after the paper's ablation: run the student with the lesson removed (question + options only) and with the question removed (lesson + options only). (Cosmos QA: Machine Reading Comprehension with Contextual… · permanent course controls: no lesson, questions shuffled, key lines removed; all red in all seven courses (biology real 2238/4581, no lesson 1208, shuffled…)

COMET-ATOMIC 2020: On Symbolic and Neural Commonsense Knowledge Graphs (2020) measured
The paper argues that hand-built commonsense knowledge graphs can never cover everything an NLP system will meet.
Cosmos QA: Machine Reading Comprehension with Contextual Commonsense Reasoning (2019) 1 idea kept
Cosmos QA has 35,588 four-option questions on 21,886 paragraphs from personal-narrative blogs, written by crowd workers to need commonsense «between the lines»: likely causes of events, what happened before or after, plausible facts about people and things, and counterfactuals («What would have happened if…?»).
Abductive Commonsense Reasoning (2019)
The paper introduces abductive reasoning over narratives. Given two observations O1 and O2, a system must pick the more plausible explanation from two hypotheses (αNLI), or generate one (αNLG). The ART dataset has about 20k contexts taken from ROCStories and over 200k hypotheses, written by crowdworkers and then…