AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Rethink Representation Learning for Questionnaire Data

Guanhua Ye, Jifeng He, Yan Li, Junping Du, Zhe Xue, Yingxia Shao, Meiyu Liang, Yawen Li

AAAI 2026 · Large language models › Language, safety and interpretability · approach: Linguistic encoding in LLM representations · 1 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 semantic-wise components (intro and terms, option text, question text) (ablation)SemantiQ · BRFSS, NLSY, MHPaccuracy change (%): -1.97, -3.73, -3.70Improved
add feature-wise components (TTT, left-out masking) (ablation)SemantiQ · BRFSS, NLSY, MHPaccuracy change (%): -1.76, -0.93, -2.54Improved
add sample-wise ratio-based augmentation (ablation)SemantiQ · BRFSS, NLSY, MHPaccuracy change (%): -1.60, -1.45, -1.10Improved
add semantic-wise and feature-wise groups together (ablation)SemantiQ · BRFSS, NLSY, MHPaccuracy change (%): -3.92, -4.89, -6.55Improved
add all three component groups (ablation)SemantiQ · BRFSS, NLSY, MHPaccuracy change (%): -5.60, -6.42, -7.71Improved
use cell-level SemantiQ instead of row-level TaBERT under column shufflingTaBERT · BRFSS, NLSY, MHPaccuracy change under shuffled order (%): -6.73 -> -0.98, -5.91 -> -0.31, -7.13 -> -0.79Improved
use cell-level SemantiQ instead of row-level TabLLM under column shufflingTabLLM · BRFSS, NLSY, MHPaccuracy change under shuffled order (%): -4.31 -> -0.98, -7.31 -> -0.31, -2.47 -> -0.79Improved
increase embedding dimension from 16 to 512 (ablation)SemantiQ · BRFSSaccuracy: about 2.3% improvementImproved
extend LLM statement length beyond 100-150 tokens (ablation)SemantiQ · BRFSS, NLSY, MHPaccuracyHurt
apply SemantiQ augmentation to XGBoost (SemantiQ-DT)XGBoost · BRFSS, NLSY, MHPaccuracy (%): 65.02 -> 69.42, 59.04 -> 63.77, 82.05 -> 90.23Improved
add semantic upsampling and SemantiQ to FT-Transformer baseline on MHPFT-Transformer · MHP (5-fold CV)accuracy (%): 94.01 -> 96.63Improved
replace row-level LLM calls with cell-level statementsTabLLM/TaBERT · BRFSSLLM calls: 401,958 -> 107 LLM callsImproved

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