Research map / Vision-language and multimodal / Visual question answering
WebQA: Multihop and Multimodal QA
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 VinVL visual features instead of standard VLP visual features (ablation) | VLP · WebQA restricted | FL×Acc: 22.6 -> 24.1 | Helped |
| Provide gold sources rather than predicted sources (ablation) | VLP · WebQA restricted | FL×Acc: 22.6 -> 24.5 | Helped |
| Add source context to the question-only prompt (ablation) | PICa · WebQA restricted, oracle sources | FL×Acc: 28.8 -> 40.1 | Helped |
| Rerank the top 20 CLIP candidates with VLP | CLIP full-scale retrieval · WebQA full image retrieval | F1: 9.71 -> 21.68 | Helped |
| Rerank the top 20 CLIP candidates with VLP | CLIP full-scale retrieval · WebQA full text retrieval | F1: 13.96 -> 26.01 | Helped |
| Use VLP rather than BM25 for candidate-source retrieval | WebQA restricted retrieval · Image sources | F1: 25.61 -> 68.13 | Helped |
| Use VLP rather than BM25 for candidate-source retrieval | WebQA restricted retrieval · Text sources | F1: 43.75 -> 69.48 | Helped |
| Add source context to the question-only model (ablation) | VLP · WebQA text QA with oracle sources | accuracy: 18.1 -> 50.4 | Helped |
| Add source context to the question-only model (ablation) | PICa · WebQA text QA with oracle sources | accuracy: 42.8 -> 74.8 | Helped |
| Add source context to the question-only model (ablation) | VLP · WebQA image Y/N QA with oracle sources | accuracy: 16.1 -> 17.2 | No clear effect |
| Add source context to the question-only model (ablation) | PICa · WebQA image Y/N QA with oracle sources | accuracy: 26.7 -> 27.4 | No clear effect |
| Scale retrieval from the restricted candidate set to the full corpus | VLP-based retrieval · WebQA source retrieval | F1 | Hurt |
Closest papers by meaning
- ReAG: Reasoning-Augmented Generation for Knowledge-based Visual Question Answering (CVPR 2026 · similarity 0.92)
- VTQA: Visual Text Question Answering via Entity Alignment and Cross-Media Reasoning (CVPR 2024 · similarity 0.91)
- When Open-Vocabulary Visual Question Answering Meets Causal Adapter: Benchmark and Approach (AAAI 2025 · similarity 0.90)
- CVQA: Culturally-diverse Multilingual Visual Question Answering Benchmark (NeurIPS 2024 · similarity 0.90)
- Separation of Powers: On Segregating Knowledge from Observation in LLM-enabled Knowledge-based Visual Question Answering (CVPR 2025 · similarity 0.90)
- BOK-VQA: Bilingual outside Knowledge-Based Visual Question Answering via Graph Representation Pretraining (AAAI 2024 · similarity 0.90)
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.