Research map / Large language models / Language, safety and interpretability
A Massive Scale Semantic Similarity Dataset of Historical English
Emily Silcock, Abhishek Arora, Melissa Dell
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use contrastive neural bi-encoder instead of LSH | reproduced content detection · NEWS-COPY test | adjusted Rand index: 73.7 -> 91.5 | Improved |
| use neural bi-encoder instead of N-gram overlap | reproduced content detection · NEWS-COPY test | adjusted Rand index: 75.0 -> 91.5 | Improved |
| add cross-encoder re-ranking on best bi-encoder matches | bi-encoder · NEWS-COPY test | adjusted Rand index: 91.5 -> 93.7 | Improved |
| add RoBERTa cross-encoder to rule-based article association to raise recall | article association · 214 hand-labeled scans | F1 | Improved |
| use hard-negative-heavy imbalanced training set (one twelfth positives, one twelfth hard negatives) (ablation) | article association cross-encoder · training data | performance | Improved |
| OCR on later decades | Tesseract OCR · HEADLINES decades | character error rate: 4.3 -> 1.5 | Improved |
| compare HEADLINES clusters with MTEB clustering datasets | language models · clustering v-measure | v-measure: 43.69 -> 78 | Improved |
Closest papers by meaning
- Newswire: A Large-Scale Structured Database of a Century of Historical News (NeurIPS 2024 · similarity 0.88)
- Style-transfer and Paraphrase: Looking for a Sensible Semantic Similarity Metric (AAAI 2021 · similarity 0.85)
- Semantic-KG: Using Knowledge Graphs to Construct Benchmarks for Measuring Semantic Similarity (NeurIPS 2025 · similarity 0.85)
- WikiSQE: A Large-Scale Dataset for Sentence Quality Estimation in Wikipedia (AAAI 2024 · similarity 0.84)
- SoftMatcha: A Soft and Fast Pattern Matcher for Billion-Scale Corpus Searches (ICLR 2025 · similarity 0.83)
- Sparse Autoencoders for Hypothesis Generation (ICML 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
