Research map / Label-efficient and robust learning / Recognition and long-tail
SensEmBERT: Context-Enhanced Sense Embeddings for Multilingual Word Sense Disambiguation
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 |
|---|---|---|---|
| knowledge-based sense embeddings from BERT on Wikipedia contexts | knowledge-based WSD · English ALL (nouns) | F1: 69.8 -> 75.9 | Helped |
| add SemCor supervision to SensEmBERT | SensEmBERT · English ALL (nouns) | F1: 78.0 -> 80.4 | Helped |
| add gloss embeddings to context embeddings (concatenation) (ablation) | SensEmBERT · ALL nouns | F1: 74.5 -> 75.9 | Helped |
| concatenate instead of average context and gloss (ablation) | SensEmBERT · ALL nouns | F1: 75.3 -> 75.9 | Helped |
| add Wikipedia-derived contexts to SemCor vectors (ablation) | SensEmBERT_sup · ALL nouns | F1: 69.3 -> 79.2 | Helped |
| add Wikipedia-derived contexts to SemCor|Gloss (ablation) | SensEmBERT_sup · ALL nouns | F1: 78.0 -> 80.4 | Helped |
| knowledge-based sense embeddings instead of SemCor-based LMMS | WSD · ALL_LFS rare senses | F1: 66.7 -> 76.9 | Helped |
| SensEmBERT unsupervised vs LMMS supervised | WSD · English ALL (nouns) | F1: 78.0 -> 75.9 | Hurt |
| SensEmBERT vs LMMS on silver OneSeC data | multilingual WSD · SemEval-13/15 multilingual | F1: on average 6.6 points | Helped |
Closest papers by meaning
- XL-WSD: An Extra-Large and Cross-Lingual Evaluation Framework for Word Sense Disambiguation (AAAI 2021 · similarity 0.92)
- Knowledge-based Word Sense Disambiguation using Topic Models (AAAI 2018 · similarity 0.89)
- VAST: The Valence-Assessing Semantics Test for Contextualizing Language Models (AAAI 2022 · similarity 0.88)
- ConceptNet 5.5: An Open Multilingual Graph of General Knowledge (AAAI 2017 · similarity 0.88)
- Learning Semantic Representations for Novel Words: Leveraging Both Form and Context (AAAI 2019 · similarity 0.88)
- Two Knowledge-based Methods for High-Performance Sense Distribution Learning (AAAI 2018 · similarity 0.88)
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.