Summarize-then-Answer: Generating Concise Explanations for Multi-hop Reading Comprehension
release_mdiqqhpitjesrap3ilgwld5rnu
by
Naoya Inoue, Harsh Trivedi, Steven Sinha, Niranjan Balasubramanian, Kentaro Inui
2021
Abstract
How can we generate concise explanations for multi-hop Reading Comprehension
(RC)? The current strategies of identifying supporting sentences can be seen as
an extractive question-focused summarization of the input text. However, these
extractive explanations are not necessarily concise i.e. not minimally
sufficient for answering a question. Instead, we advocate for an abstractive
approach, where we propose to generate a question-focused, abstractive summary
of input paragraphs and then feed it to an RC system. Given a limited amount of
human-annotated abstractive explanations, we train the abstractive explainer in
a semi-supervised manner, where we start from the supervised model and then
train it further through trial and error maximizing a conciseness-promoted
reward function. Our experiments demonstrate that the proposed abstractive
explainer can generate more compact explanations than an extractive explainer
with limited supervision (only 2k instances) while maintaining sufficiency.
In text/plain
format
Archived Files and Locations
application/pdf 947.1 kB
file_utvweobsobbq5hhhg42pef6m6q
|
arxiv.org (repository) web.archive.org (webarchive) |
2109.06853v1
access all versions, variants, and formats of this works (eg, pre-prints)