Reproduce “Finding your MUSE: Mining Unexpected Solutions Engine” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Nir Sweed, Hanit Hakim, Ben Wolfson, Hila Lifshitz, Dafna Shahaf
Researchers from The Hebrew University of Jerusalem developed a scalable methodology using Large Language Models to construct a high-quality Functional Concept Graph (FCG) from 500,000 patents. Their Mining Unexpected Solutions Engine (MUSE) algorithm, which navigates this FCG to generate diverse inspirations, significantly increases human creativity by up to 75% in problem-solving tasks, particularly when inspirations are presented with clear contextualized sentences.
Key insights
- Providing inspirations in a contextualized 'Purpose + mechanism sentence' format is most effective, leading to a higher ratio and absolute number of creative ideas compared to simpler formats.
- Inspirations derived from NLI-based and LLM-based abstract connections within the FCG are more consistently effective at fostering creative and feasible solutions than purely verb-based connections.
- While initially requiring more processing, structured AI-generated inspirations ultimately lead to higher-quality idea generation over time, demonstrating a synergistic effect between human and AI cognitive processes.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.