Research map / Label-efficient and robust learning / Recognition and long-tail
Discriminative Sentence Modeling for Story Ending Prediction
Yiming Cui, Wanxiang Che, Wei-Nan Zhang, Ting Liu, Shijin Wang, Guoping Hu
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 the Diff-Net discriminative module | BERT · SCT v1.0 test | accuracy: 89.2 -> 90.1 | Improved |
| add Diff-Net to BERT | BERT · SCT v1.5 test | accuracy: 80.7 -> 82.0 | Improved |
| add BERT representations as features | Diff-Net · SCT v1.0 test | accuracy: 77.8 -> 84.3 | Improved |
| add cosine loss to separate ending representations (ablation) | Diff-Net · SCT v1.0 test | accuracy: 77.2 ($77.02 \pm 0.20$) -> 77.8 ($77.60 \pm 0.12$) | Improved |
| use modified AoA (ablation) | Diff-Net · SCT v1.0 test | accuracy: 77.0 ($76.93 \pm 0.07$) -> 77.8 ($77.60 \pm 0.12$) | Improved |
| use the match module to retrieve story-relevant information (ablation) | Diff-Net · SCT v1.0 test | accuracy: 77.1 ($76.85 \pm 0.20$) -> 77.8 ($77.60 \pm 0.12$) | Improved |
| use the discriminative module to compare endings (ablation) | Diff-Net · SCT v1.0 test | accuracy: 76.9 ($76.75 \pm 0.21$) -> 77.8 ($77.60 \pm 0.12$) | Improved |
| use AoA instead of a simple dot product (ablation) | Diff-Net · SCT v1.0 test | accuracy: 76.7 ($76.58 \pm 0.11$) -> 77.8 ($77.60 \pm 0.12$) | Improved |
| use binary matching features (ablation) | Diff-Net · SCT v1.0 test | accuracy: 76.0 ($75.78 \pm 0.12$) -> 77.8 ($77.60 \pm 0.12$) | Improved |
| add End-End Match (ablation) | Diff-Net · SCT v1.0 test | accuracy: 77.2 ($77.05 \pm 0.11$) -> 77.8 ($77.60 \pm 0.12$) | Improved |
| add End-Story Match (ablation) | Diff-Net · SCT v1.0 test | accuracy: 77.0 ($76.83 \pm 0.12$) -> 77.8 ($77.60 \pm 0.12$) | Improved |
| add End-Story Fuzzy Match (ablation) | Diff-Net · SCT v1.0 test | accuracy: 76.9 ($76.78 \pm 0.11$) -> 77.8 ($77.60 \pm 0.12$) | Improved |
Closest papers by meaning
- Incorporating Structured Commonsense Knowledge in Story Completion (AAAI 2019 · similarity 0.89)
- Story Ending Generation with Multi-Level Graph Convolutional Networks over Dependency Trees (AAAI 2021 · similarity 0.86)
- What Happens Next? Event Prediction Using a Compositional Neural Network Model (AAAI 2016 · similarity 0.85)
- Show Me a Story: Towards Coherent Neural Story Illustration (CVPR 2018 · similarity 0.85)
- Story Ending Generation with Incremental Encoding and Commonsense Knowledge (AAAI 2019 · similarity 0.84)
- StoryGPT-V: Large Language Models as Consistent Story Visualizers (CVPR 2025 · similarity 0.80)
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.
