Problem
Research questions and friction points this paper is trying to address.
scientific problems
knowledge base
problem-solution-rationale
cross-domain
full-text mining
MUSE: A Full-Text Cross-Domain Knowledge Base of Scientific Problems, Solutions, and Rationales
🤖 AI Summary
Scientific literature contains rich, fine-grained records of problem-solving processes, yet lacks a structured, cross-domain knowledge base for systematically extracting problems, their solutions, and the underlying reasoning. This work proposes MUSE, the first multidisciplinary knowledge base of problem-solution-reasoning (P-S-R) triples, constructed through expert annotation and a modular extraction pipeline that yields 37K high-quality, context-anchored triples from full-text articles. The resource incorporates enriched annotations such as semantic links and conceptual coreference. Experimental results demonstrate that supervised reasoning based on this knowledge base significantly enhances large language models’ performance on complex, multi-constraint problems, validating the efficacy of explicit reasoning modeling—albeit with a slight performance trade-off on simpler tasks.
📝 Abstract
Scientific papers contain fine-grained records of problem solving: authors mention technical obstacles and methods that were used to address them, often along with reasoning on why those methods were chosen. We introduce MUSE (Mining Underlying Scientific Explanations), a full-text, multi-domain resource of scientific Problem-Solution-Rationale (P-S-R) triplets. We curate 579 expert-annotated full-text paragraphs, with a rich annotation schema covering salient problem, solution, and rationale spans, solves and rationale_of links and conceptual coreference. A modular extraction pipeline scales this annotation to build a high-quality knowledge base of 37K source-grounded P-S-R triplets. We evaluate the extraction components and include a preliminary experiment training a rationale-supervised LLM for scientific problem solving. Interestingly, we find that rationale supervision improves performance on complex, multi-constraint problems but can harm performance on simpler ones.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Problem-Solution-Rationale extraction
scientific knowledge base
rationale-supervised learning
cross-domain full-text mining
modular information extraction pipeline
🔎 Similar Papers
No similar papers found.
💼 Related Jobs
No related jobs found.
