Research map / Large language models / Agents, code and math
EMODIS: A Benchmark for Context-Dependent Emoji Disambiguation in Large Language Models
Jiacheng Huang, Ning Yu, Xiaoyin Yi
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 API-based GPT-4 instead of open-source models | EMODIS emoji disambiguation · EMODIS overall | pair-wise accuracy (Accp, %): 58.8 (GPT-4) vs 34.7 (Qwen-14B) | Improved |
| scale LLaMA from 7B to 13B | LLaMA · EMODIS overall | pair-wise accuracy (Accp, %): 13.5 -> 21.3 | Improved |
| scale Qwen from 7B to 14B | Qwen · EMODIS overall | pair-wise accuracy (Accp, %): 26.5 -> 34.7 | Improved |
| use social intent contexts instead of temporal contexts | GPT-4 · EMODIS | pair-wise accuracy (Accp, %): 64.5 -> 54.2 | Hurt |
| use open-source models (LLaMA-7B) instead of GPT-4 on context awareness | context sensitivity · EMODIS | Score_ca: 0.406 -> 0.254 | Hurt |
| replace human with best LLM on context awareness | context sensitivity · EMODIS | Score_ca: 0.959 -> 0.406 | Hurt |
| swap GPT-4 judge for human judge (ablation) | EMODIS evaluation · EMODIS | Accp (%): 59.6 (GPT-4 judge) vs 60.1 (human judge) for GPT-4 outputs | No clear effect |
Closest papers by meaning
- Quantifying the Plausibility of Context Reliance in Neural Machine Translation (ICLR 2024 · similarity 0.84)
- Emoji Attack: Enhancing Jailbreak Attacks Against Judge LLM Detection (ICML 2025 · similarity 0.83)
- ExLM: Rethinking the Impact of $\texttt{[MASK]}$ Tokens in Masked Language Models (ICML 2025 · similarity 0.83)
- ImpScore: A Learnable Metric For Quantifying The Implicitness Level of Sentences (ICLR 2025 · similarity 0.83)
- NoLiMa: Long-Context Evaluation Beyond Literal Matching (ICML 2025 · similarity 0.82)
- Less is More: Local Intrinsic Dimensions of Contextual Language Models (NeurIPS 2025 · similarity 0.82)
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.
