Research map / Label-efficient and robust learning / Recognition and long-tail
Improving Twitter Sentiment Classification Using Topic-Enriched Multi-Prototype Word Embeddings
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 |
|---|---|---|---|
| add topic prediction loss to C&W embeddings (ablation) | C&W · SemEval 2013 | macro-F: 67.75 -> 82.17 | Helped |
| add topic loss to sentiment-specific embeddings (ablation) | SSWE · SemEval 2013 | macro-F: 84.98 -> 85.34 | Helped |
| use multi-prototype embeddings (ablation) | TSWE · SemEval 2013 | macro-F: 85.34 -> 86.83 | Helped |
| use multi-prototype embeddings (ablation) | C&W · SemEval 2013 | macro-F: 67.75 -> 72.37 | Helped |
| add topic loss to multi-prototype sentiment embeddings (ablation) | M-SSWE · SemEval 2013 | macro-F: 86.10 -> 86.83 | Helped |
| ensemble with NRC hand-crafted system | M-TSWE · SemEval 2013 | macro-F: 86.83 -> 87.97 | Helped |
| set sentiment and topic loss weights to 0 (n-gram only) (ablation) | M-TSWE · SemEval 2013 dev | macro-F | Hurt |
| set alpha=beta=0.5 removing n-gram loss (ablation) | M-TSWE · SemEval 2013 dev | macro-F | Hurt |
| vary LDA topic number 10 to 30 (ablation) | M-TSWE · SemEval 2013 dev | macro-F | Mixed |
| add topic loss to embeddings | SSWE · HL lexicon | polarity consistency accuracy: 77.18 -> 79.96 | Helped |
Closest papers by meaning
- Context-Sensitive Twitter Sentiment Classification Using Neural Network (AAAI 2016 · similarity 0.87)
- Enriching Word Embeddings with a Regressor Instead of Labeled Corpora (AAAI 2019 · similarity 0.87)
- Learning Sentiment-Specific Word Embedding via Global Sentiment Representation (AAAI 2018 · similarity 0.85)
- SNNN: Promoting Word Sentiment and Negation in Neural Sentiment Classification (AAAI 2018 · similarity 0.84)
- Representing Mixtures of Word Embeddings with Mixtures of Topic Embeddings (ICLR 2022 · similarity 0.83)
- Training and Evaluating Multimodal Word Embeddings with Large-scale Web Annotated Images (NeurIPS 2016 · similarity 0.83)
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.