Reproduce “StructSim: Measuring Idea Similarity at Scale Through Structural Representation” 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
Yaqing Yang, Vikram Mohanty, Mei-Xi Chia, Nikolas Martelaro, Dafna Shahaf, Aniket Kittur
StructSim measures idea similarity by decomposing each idea into purpose, mechanism, and implementation, separating core from supporting parts and comparing shared concepts and their links. In expert comparisons, it matched human-majority choices with 0.871 accuracy, versus 0.562 for the strongest embedding baseline, among cases where experts chose one candidate rather than “neither.” Its evidence is limited to tested design tasks.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.