Papers by Zhiguo Gong
Self-distilled Transitive Instance Weighting for Denoised Distantly Supervised Relation Extraction (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing approaches to reducing wrongly labeled instances are based on a bag-level setting . however, sentence-level training is vulnerable to the noise brought by DS, which limits its application. |
| Approach: | They propose a transitive instance weighting mechanism integrated with the self-distilled BERT backbone to generate dynamic instance weights for denoised sentence-level training. |
| Outcome: | The proposed method can tackle wrongly labeled instances and prevent overfitting. |
Distantly Supervised Relation Extraction using Multi-Layer Revision Network and Confidence-based Multi-Instance Learning (2021.emnlp-main)
Copied to clipboard
| Challenge: | Distantly supervised relation extraction is used in knowledge bases but its low quality and noisy sentences are present in sentence bags. |
| Approach: | They propose a multi-layer revision network which emphasizes inner-sentence correlations before extracting relevant information within sentences. |
| Outcome: | The proposed method improves on two New York Times datasets. |
KVFKT: A New Horizon in Knowledge Tracing with Attention-Based Embedding and Forgetting Curve Integration (2025.coling-main)
Copied to clipboard
| Challenge: | Existing knowledge tracing models do not incorporate forgetting features to improve the learning and answering processes. |
| Approach: | They propose a new approach in knowledge tracing with attention-based embedding and forgetting curve integration using four real-world datasets to test the model. |
| Outcome: | The proposed model outperforms the existing knowledge tracing models and eliminates the need for artificial engineering features. |