Reproduce “Explainable Semantic Textual Similarity via Dissimilar Span Detection” 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
Diego Miguel Lozano, Daryna Dementieva, Alexander Fraser
The authors introduce Dissimilar Span Detection (DSD), which marks corresponding text spans with different meanings alongside a sentence-similarity score, and release the Span Similarity Dataset. Their replacement-based Embedding-DSD method detects candidate spans by measuring similarity changes. On PAWS-Wiki, adding a no-dissimilar-span condition raises test accuracy from 0.795 to 0.868; this is one setup, and the methods remain baselines.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.