Research map / Label-efficient and robust learning / Recognition and long-tail
Single or Multiple? Combining Word Representations Independently Learned from Text and WordNet
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 |
|---|---|---|---|
| combine independently learned WordNet embeddings with text embeddings via concatenation | WBU skipgram · all-dataset average | Spearman: 64.5 -> +8.4 gain | Helped |
| reduce concatenation to 300d with PCA | CAT(WBU+RWwn) · all-dataset average | Spearman gain: +8.4 -> +8.9 | Helped |
| use complex-number combination instead of concatenation (ablation) | WBU+RWwn · all-dataset average | Spearman gain: -4.0 vs +8.4 for CAT | Hurt |
| use rank averaging instead of concatenation (ablation) | WBU+RWwn · all-dataset average | Spearman gain: -0.8 vs +8.4 for CAT | Hurt |
| use CCA instead of concatenation (ablation) | WBU+RWwn · all-dataset average | Spearman gain: +2.3 vs +8.4 for CAT | No clear effect |
| combine with independent WordNet embeddings instead of retrofitting WordNet into WBU | WBU · all-dataset average | Spearman gain: +2.1 (retrofit WNall) -> +8.9 (PCA) | Helped |
| increase from 2-way to 3/4-way concatenation (ablation) | CAT · all-dataset average | Spearman: 72.9 -> 72.2 | Hurt |
| increase from 2-way to 6-way averaging | AVG · all-dataset average | Spearman: 72.7 -> 75.7 | Helped |
| remove RWwn from six-way AVG (ablation) | AVG(ALL) · all-dataset average | Spearman: 73.3 vs 75.7 (adding RWwn) | Helped |
| remove WBU from six-way AVG (ablation) | AVG(ALL) · all-dataset average | Spearman: 75.5 vs 75.7 (adding WBU) | Helped |
Closest papers by meaning
- Joint Word Representation Learning Using a Corpus and a Semantic Lexicon (AAAI 2016 · similarity 0.90)
- Semantic Structure-Based Word Embedding by Incorporating Concept Convergence and Word Divergence (AAAI 2018 · similarity 0.88)
- Leveraging Web Semantic Knowledge in Word Representation Learning (AAAI 2019 · similarity 0.87)
- Understanding the Semantic Content of Sparse Word Embeddings Using a Commonsense Knowledge Base (AAAI 2020 · similarity 0.87)
- ConceptNet 5.5: An Open Multilingual Graph of General Knowledge (AAAI 2017 · similarity 0.87)
- Unseen Word Representation by Aligning Heterogeneous Lexical Semantic Spaces (AAAI 2019 · similarity 0.87)
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.