Research map / Image generation / Editing and style transfer
PPTArena: A Benchmark for PowerPoint Editing
Michael Ofengenden, Yunze Man, Ziqi Pang, Liang-Yan Gui, Yu-Xiong Wang
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 |
|---|---|---|---|
| add skill routing between XML and python-pptx (vs XML-only) (ablation) | PPTPilot · PPTArena | IF: 0.95 -> 2.36 | Improved |
| add skill routing (vs python-pptx-only) (ablation) | PPTPilot · PPTArena | IF: 2.06 -> 2.36 | Improved |
| add 3x iterative reflection loop (ablation) | PPTPilot hybrid · PPTArena | IF: 2.36 -> 2.84 | Improved |
| add 3x iterative reflection loop (ablation) | PPTPilot hybrid · PPTArena | VQ: 2.69 -> 3.21 | Improved |
| use dual judge with structured diffs instead of single judge (ablation) | VLM judge · PPTArena | VQ: 4.26 -> 2.40 | Improved |
| drop structured diffs from dual judge (ablation) | VLM judge · PPTArena | IF: 2.36 -> 3.76 (inflated) | Hurt |
| use PPTPilot instead of ChatGPT | agent · PPTArena full | IF: 2.07 -> 2.57 | Improved |
| use PPTPilot instead of Gemini CLI | agent · PPTArena full | VQ: 1.98 -> 2.69 | Improved |
| use PPTPilot instead of ChatGPT Agent on Structure | agent · PPTArena hard subset | IF: 1.47 vs 2.00 | Hurt |
| swap judge to Gemini 3.1 Pro | PPTPilot ranking · PPTArena full | IF: 2.57 -> 2.45 | No clear effect |
| use PPTPilot instead of T2US method | agent · T2US benchmark | IF: 2.21 -> 4.05 | Improved |
Closest papers by meaning
- PPT-Eval: A Benchmark for Computer-Use Agents on PowerPoint Tasks (ICML 2026 · similarity 0.91)
- EdiVal-Agent: An Object-Centric Framework for Automated, Fine-Grained Evaluation of Multi-Turn Editing (ICLR 2026 · similarity 0.86)
- AutoPresent: Designing Structured Visuals from Scratch (CVPR 2025 · similarity 0.86)
- KRIS-Bench: Benchmarking Next-Level Intelligent Image Editing Models (NeurIPS 2025 · similarity 0.85)
- UnicEdit-10M: A Dataset and Benchmark Breaking the Scale-Quality Barrier via Unified Verification for Reasoning-Enriched Edits (CVPR 2026 · similarity 0.84)
- ImgEdit: A Unified Image Editing Dataset and Benchmark (NeurIPS 2025 · similarity 0.84)
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.
