AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

A Generalized Idiom Usage Recognition Model Based on Semantic Compatibility

Changsheng Liu, Rebecca Hwa

AAAI 2019 · Label-efficient and robust learning › Recognition and long-tail · approach: Multilingual language representation pretraining

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 changedWhereMeasuredEffect
Remove semantic compatibility layer (ablation)ACE+LocalAtt+AKEW · SemEval+VNC averageavg figurative F-score: 0.76 -> 0.67Hurt
Remove semantic compatibility layer (ablation)ACE+GlobalAtt+AKEW · SemEval+VNC averageavg figurative F-score: 0.74 -> 0.66Hurt
Remove semantic compatibility layer (ablation)BiLSTM+GlobalAtt+AKEW · SemEval+VNC averageavg figurative F-score: 0.71 -> 0.65Hurt
Remove semantic compatibility layer (ablation)BiLSTM+LocalAtt+AKEW · SemEval+VNC averageavg figurative F-score: 0.73 -> 0.66Hurt
Remove attention (ablation)ACE+AKEW · SemEval+VNC averageavg figurative F-score: 0.76 -> 0.66Hurt
Remove attention (ablation)BiLSTM+AKEW · SemEval+VNC averageavg figurative F-score: 0.73 -> 0.7Hurt
Use local attention instead of global attention (ablation)BiLSTM+AKEW · SemEval+VNC averageavg figurative F-score: 0.71 -> 0.73Improved
Use BiLSTM instead of ACE without attention (ablation)AKEW · SemEval+VNC averageavg figurative F-score: 0.66 -> 0.7Improved
Use BiLSTM instead of ACE with attention (ablation)LocalAtt+AKEW · SemEval+VNC averageavg figurative F-score: 0.76 -> 0.73Hurt
Use AKWE instead of AWE for idiom literal representation (ablation)ACE+LocalAtt · SemEvalavg figurative F-score: 0.74 -> 0.76Improved
Use AKWE instead of AWE for idiom literal representation (ablation)ACE+GlobalAtt · SemEvalavg figurative F-score: 0.72 -> 0.74Improved
Use AKWE instead of AWE (ablation)ACE+LocalAtt · VNCavg figurative F-score: 0.76 -> 0.75Hurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy