Research map / Video / Video-language understanding
MovieQA: Understanding Stories in Movies Through Question-Answering
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 plot synopses rather than subtitles, scripts, or DVS as the text story | MovieQA text QA systems · MovieQA val | accuracy | Helped |
| Fuse text representations with SSCB rather than use TF-IDF alone | SSCB · MovieQA val, Plot | accuracy: 48.5 -> 56.7 | Helped |
| Use the learned SSCB scoring model instead of cosine similarity | TF-IDF story matching · MovieQA val, Plot | accuracy: 47.6 -> 48.5 | Helped |
| Use MemN2N rather than SSCB for long text sources | MovieQA text QA · MovieQA val, DVS, Subtitle, and Script | accuracy: 24.5 -> 33.0; 27.6 -> 38.0; 26.1 -> 42.3 | Helped |
| Replace learned MemN2N word embeddings with fixed Word2Vec embeddings and a shared linear projection (ablation) | MemN2N · MovieQA val, Plot | accuracy: near random performance on val (~20%) -> 40.6 | Helped |
| Use video clips rather than subtitles | MovieQA video QA · MovieQA val | accuracy: 38.0 -> 23.1; 22.3 -> 21.6 | Hurt |
| Late-fuse video and subtitle scores | SSCB · MovieQA val, Video+Subtitle | accuracy: 21.6 -> 21.9 | No clear effect |
| Choose the longest answer without story evidence | Hasty Student · MovieQA val | accuracy: 20% -> 25.33 | Helped |
| Choose the most distinct answer by its representation | Hasty Student · MovieQA val | accuracy: 19.92; 14.91; 15.12 | Hurt |
| Answer without access to the story | Human turkers · MovieQA subset | accuracy: 27.6% -> 24.7% | Mixed |
Closest papers by meaning
- FriendsQA: A New Large-Scale Deep Video Understanding Dataset with Fine-grained Topic Categorization for Story Videos (AAAI 2025 · similarity 0.90)
- LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understanding (NeurIPS 2024 · similarity 0.85)
- Understanding Complexity in VideoQA via Visual Program Generation (ICML 2025 · similarity 0.85)
- CG-Bench: Clue-grounded Question Answering Benchmark for Long Video Understanding (ICLR 2025 · similarity 0.85)
- VTQA: Visual Text Question Answering via Entity Alignment and Cross-Media Reasoning (CVPR 2024 · similarity 0.85)
- MoReVQA: Exploring Modular Reasoning Models for Video Question Answering (CVPR 2024 · similarity 0.85)
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.