Research map / Probabilistic and scientific modeling / Causal discovery
Distributed Representation of Words in Cause and Effect Spaces
Zhipeng Xie, Feiteng Mu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| use Max-Matching (one positive word pair per positive phrase pair) instead of Pairwise-Matching (ablation) | causal embedding · Sogou test (trained on Baike) | accuracy: 24.9% -> 58.6% | Improved |
| use Max-Matching instead of Pairwise-Matching (ablation) | causal embedding · Baidu test (trained on Sogou) | accuracy: 19.5% -> 42.9% | Improved |
| use Attentive-Matching instead of Pairwise-Matching (ablation) | causal embedding · Sogou test (trained on Baike) | accuracy: 24.9% -> 53.0% | Improved |
| use Attentive-Matching instead of Pairwise-Matching (ablation) | causal embedding · Baidu test (trained on Sogou) | MRR: 0.348 -> 0.572 | Improved |
| Attentive-Matching (soft) instead of Max-Matching (hard) (ablation) | causal embedding · Sogou test (trained on Baike) | accuracy: 58.6% -> 53.0% | Hurt |
| Attentive-Matching instead of Max-Matching (ablation) | causal embedding · Baidu test (trained on Sogou) | accuracy: 42.9% -> 42.1% | No clear effect |
| Pairwise-Matching vs cEmbed (all word pairs assumed causal) (ablation) | causal embedding · Sogou test (trained on Baike) | accuracy: 19.1% -> 24.9% | No clear effect |
| bidirectional cEmbedBi vs unidirectional cEmbed (ablation) | cEmbed · Sogou test (trained on Baike) | accuracy: 19.1% -> 23.5% | Improved |
| noise-aware weighting (cEmbedBiNoise) vs cEmbedBi (ablation) | cEmbedBi · Baidu test (trained on Sogou) | accuracy: 19.2% -> 18.3% | Hurt |
| causal embedding instead of vanilla embedding (vEmbed) | vEmbed · Baidu test (trained on Sogou) | accuracy: 8.4% -> 42.9% | Improved |
| add all causal embedding features to CR+vEmbed reranker | CR+vEmbed · Causal QA (Yahoo! Answers) | P@1: 34.6 -> 37.9 | Improved |
| add vanilla embedding features to candidate retrieval | CR · Causal QA (Yahoo! Answers) | P@1: 24.3 -> 34.6 | Improved |
Closest papers by meaning
- Understanding the Semantic Content of Sparse Word Embeddings Using a Commonsense Knowledge Base (AAAI 2020 · similarity 0.87)
- A Simple but Tough-to-Beat Baseline for Sentence Embeddings (ICLR 2017 · similarity 0.86)
- VAST: The Valence-Assessing Semantics Test for Contextualizing Language Models (AAAI 2022 · similarity 0.86)
- Causally Denoise Word Embeddings Using Half-Sibling Regression (AAAI 2020 · similarity 0.86)
- Dynamic Word Embeddings (ICML 2017 · similarity 0.86)
- By Tying Embeddings You Are Assuming the Distributional Hypothesis (ICML 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.
