Research map / Graph learning / GNN architectures
Graph-Based Tri-Attention Network for Answer Ranking in CQA
Read the paperCheck 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 GTAN instead of AUANN | CQA answer ranking · StackOverflow | P@1: 40.66% to 42.30% | Helped |
| use GTAN instead of AUANN | CQA answer ranking · Zhihu | NDCG@3: 59.58% to 61.34% | Helped |
| use GTAN instead of AUANN | CQA answer ranking · Quora | MRR: 81.52% to 83.68% | Helped |
| add CQA-based graph message passing to model answer correlations (ablation) | GTAN · StackOverflow, Zhihu, Quora | P@1: 0.3828 -> 0.4230; 0.3302 -> 0.3556; 0.7037 -> 0.7235 | Helped |
| use type-specific rather than shared transformation matrices (ablation) | GTAN graph neural network · StackOverflow, Zhihu, Quora | P@1: 0.4104 -> 0.4230; 0.3325 -> 0.3556; 0.7117 -> 0.7235 | Helped |
| include question information (ablation) | GTAN · StackOverflow, Zhihu, Quora | P@1: 0.4108 -> 0.4230; 0.3372 -> 0.3556; 0.7123 -> 0.7235 | Helped |
| include respondent information (ablation) | GTAN · StackOverflow, Zhihu, Quora | P@1: 0.3211 -> 0.4230; 0.1957 -> 0.3556; 0.4343 -> 0.7235 | Helped |
| add alternating tri-attention computations (ablation) | GTAN · StackOverflow, Zhihu, Quora | P@1: 0.4027 -> 0.4230; 0.3356 -> 0.3556; 0.7085 -> 0.7235 | Helped |
| add answer-guided, answer-specific question attention (ablation) | GTAN · StackOverflow, Zhihu, Quora | P@1: 0.4127 -> 0.4230; 0.3432 -> 0.3556; 0.7162 -> 0.7235 | Helped |
| include respondent information when attending over answer words (ablation) | GTAN answer attention · StackOverflow, Zhihu, Quora | P@1: 0.4122 -> 0.4230; 0.3363 -> 0.3556; 0.7130 -> 0.7235 | Helped |
| add QA-guided gating to form target-aware respondent representations (ablation) | GTAN · StackOverflow, Zhihu, Quora | P@1: 0.4109 -> 0.4230; 0.3470 -> 0.3556; 0.7099 -> 0.7235 | Helped |
| increase graph propagation from one layer to two (ablation) | GTAN · StackOverflow, Zhihu, Quora | ranking performance | Helped |
Closest papers by meaning
- Community-Based Question Answering via Asymmetric Multi-Faceted Ranking Network Learning (AAAI 2017 · similarity 0.89)
- Attentive Interactive Neural Networks for Answer Selection in Community Question Answering (AAAI 2017 · similarity 0.87)
- Hybrid Attentive Answer Selection in CQA With Deep Users Modelling (AAAI 2018 · similarity 0.87)
- G-Retriever: Retrieval-Augmented Generation for Textual Graph Understanding and Question Answering (NeurIPS 2024 · similarity 0.86)
- Question Calibration and Multi-Hop Modeling for Temporal Question Answering (AAAI 2024 · similarity 0.85)
- KG-Guard: Graph-Based Hallucination Detection for Knowledge Base Question Answering (NeurIPS 2026 · similarity 0.83)
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.