Research map / Label-efficient and robust learning / Recognition and long-tail
Learning Sentiment-Specific Word Embedding via Global Sentiment Representation
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 global sentiment representation (SWRank) vs sentence-window sentiment embeddings (SE-HyRank) | sentiment-specific word embedding · IMDB test (Avg features) | accuracy: 84.13 -> 88.55 | Helped |
| add global sentiment representation (SWPredict) vs SE-Pred | sentiment-specific word embedding · Twitter SemEval13 test (Avg features) | accuracy: 80.9 -> 84.05 | Helped |
| use sentence-window sentiment embeddings (SE-Pred) instead of plain Word2vec on documents | IMDB document classification · IMDB test (Avg features) | accuracy: 86.51 -> 83.62 | Hurt |
| use triplet ranking (SWRank) instead of classification (SWPredict) | SW models · IMDB test (Avg features) | accuracy: 88.31 -> 88.55 | Helped |
| use triplet ranking (SWRank) instead of classification (SWPredict) | SW models · Twitter SemEval13 test (Avg features) | accuracy: 84.05 -> 84.02 | No clear effect |
| use average of embeddings instead of max/min/avg conv concatenation as features | SWRank · Twitter SemEval13 test | accuracy: 82.69 -> 84.02 | Helped |
| use beta=1 (remove sentiment component) instead of beta 0.5-0.6 (ablation) | SWPredict/SWRank · IMDB and Twitter | accuracy | Hurt |
| set margin lambda away from 3 (IMDB) / 2 (Twitter) (ablation) | SWRank · IMDB and Twitter | accuracy | Hurt |
| use SWRank instead of Doc2VecC for document representation | IMDB · IMDB document representation | accuracy %: 88.3 -> 89.6 | Helped |
Closest papers by meaning
- A Simple but Tough-to-Beat Baseline for Sentence Embeddings (ICLR 2017 · similarity 0.89)
- Enriching Word Embeddings with a Regressor Instead of Labeled Corpora (AAAI 2019 · similarity 0.88)
- Learning Context-Specific Word/Character Embeddings (AAAI 2017 · similarity 0.88)
- Fast PMI-Based Word Embedding with Efficient Use of Unobserved Patterns (AAAI 2019 · similarity 0.87)
- No Training Required: Exploring Random Encoders for Sentence Classification (ICLR 2019 · similarity 0.87)
- Repurposing Language Models into Embedding Models: Finding the Compute-Optimal Recipe (NeurIPS 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.