Research map / Label-efficient and robust learning / Recognition and long-tail
Unseen Word Representation by Aligning Heterogeneous Lexical Semantic Spaces
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Align WordNet node embeddings with corpus embeddings using CCA | Word embeddings · SimVerb-3500, T=0 | Pearson r: 0.07 -> 0.42 | Helped |
| Use Align rather than WordNet gloss- or related-word-based induction | Unseen word embeddings · SimVerb-3500, T=0 | Pearson r: 0.28, 0.21, 0.23 -> 0.42 | Helped |
| Combine graph-derived embeddings with corpus embeddings | w2v-gn · RW | Spearman ρ: 0.45 -> 0.48 | Helped |
| Add induced embeddings for vocabulary gaps | Text classification initialized with full corpus embeddings · IMDB | accuracy: 85.4 -> 86.8 | Helped |
| Add induced embeddings for vocabulary gaps | Text classification initialized with full corpus embeddings · PL04 | accuracy: 66.2 -> 63.7 | Hurt |
| Replace embeddings for 20% of words with induced embeddings | Text classification · Ohsumed | accuracy: 16.4 -> 25.7 | Helped |
| Replace embeddings for 40% of words with induced embeddings | Text classification · Ohsumed | accuracy: 15.7 -> 28.5 | Helped |
| Use FastText subword embeddings instead of Align | Rare-word induction · MEN-3000, T=0 | Pearson r: 0.65 -> 0.76 | Helped |
| Use FastText subword embeddings instead of Align | Rare-word induction · WS-353 Sim, T=0 | Pearson r: 0.71 -> 0.74 | Helped |
| Replace CCA alignment with least-squares regression (ablation) | Alignment of KB and corpus spaces · MTURK-771 parameter experiments | performance | Hurt |
Closest papers by meaning
- Understanding the Semantic Content of Sparse Word Embeddings Using a Commonsense Knowledge Base (AAAI 2020 · similarity 0.90)
- Learning Semantic Representations for Novel Words: Leveraging Both Form and Context (AAAI 2019 · similarity 0.89)
- Semantic Structure-Based Word Embedding by Incorporating Concept Convergence and Word Divergence (AAAI 2018 · similarity 0.89)
- Repurposing Language Models into Embedding Models: Finding the Compute-Optimal Recipe (NeurIPS 2024 · similarity 0.87)
- Harnessing the Universal Geometry of Embeddings (NeurIPS 2025 · similarity 0.87)
- Select and Augment: Enhanced Dense Retrieval Knowledge Graph Augmentation (Abstract Reprint) (AAAI 2024 · similarity 0.86)
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.