AI Paper Atlas

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

Word-Level Contextual Sentiment Analysis with Interpretability

Tomoki Ito, Kota Tsubouchi, Hiroki Sakaji, Tatsuo Yamashita, Kiyoshi Izumi

AAAI 2020 · Label-efficient and robust learning › Recognition and long-tail · approach: Multilingual language representation pretraining · 15 citations (OpenAlex)

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
add lexical initialization of word sentiment from a small dictionary (ablation)SINN · WOSL, EcoRev Imacro F1: .492 -> .839Improved
add lexical initialization (ablation)SINN · LWCL, EcoRev Imacro F1: .350 -> .800Improved
add lexical initialization (ablation)SINN · GWCL, EcoRev IPearson correlation: .053 -> .588Improved
add lexical initialization (ablation)SINN · WOSL, Sentiment 140macro F1: .444 -> .737Improved
reduce dictionary words in LEXIL from 200 to 50 (ablation)SINN · WOSL, EcoRev Imacro F1: .839 -> .844No clear effect
reduce dictionary words in LEXIL from 200 to 50 (ablation)SINN · LWCL, EcoRev Imacro F1: .800 -> .776 (100 words: .815)Mixed
reduce dictionary words in LEXIL from 200 to 50 (ablation)SINN · Word-level polarity, EcoRev Imacro F1: .719 -> .695Hurt
use SINN instead of best prior GWCL attention methodGWCL · YahooPearson correlation: .405 (LBSA) vs .285 (SINN 100)Hurt
use SINN instead of IntGrad+RNNWCSA · Document-level polarity, EcoRev IImacro F1: .929 -> .944Improved

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