OneRel: Joint Entity and Relation Extraction with One Module in One Step
Authors: Yu-Ming Shang, Heyan Huang, Xianling Mao11285-11293
AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experimental results on two widely used datasets demonstrate that the proposed method performs better than the state-of-the-art baselines, and delivers consistent performance gain on complex scenarios of various overlapping patterns and multiple triples. |
| Researcher Affiliation | Academia | Yu-Ming Shang1, Heyan Huang1,2, Xian-Ling Mao1* 1School of Computer Science & Technology, Beijing Institute of Technology, Beijing, China 2Beijing Engineering Research Center of High Volume Language Information Processing and Cloud Computing Applications, Beijing, China {ymshang, hhy63, maoxl}@bit.edu.cn |
| Pseudocode | No | The paper describes the methodology and algorithms in prose and mathematical formulas, but it does not include a clearly labeled pseudocode block or algorithm section. |
| Open Source Code | No | The paper does not contain any explicit statement about open-sourcing the code for the described methodology, nor does it provide a link to a code repository. |
| Open Datasets | Yes | Following previous works (Wei et al. 2020; Wang et al. 2020; Zheng et al. 2021), we evaluate our model and all baselines on two widely used datasets: NYT (Riedel, Yao, and Mc Callum 2010) and Web NLG (Gardent et al. 2017). |
| Dataset Splits | Yes | Detailed statistics of the two datasets are described in Table 1. Train Valid Test Relations Normal SEO EPO HTO N=1 N=2 N=3 N=4 N>5 Triples NYT 56,195 4,999 5,000 24... Web NLG 5,019 500 703 171... |
| Hardware Specification | Yes | In our experiments, all training process is completed on a work station with an AMD 7742 2.25G CPU, 256G memory, a single RTX 3090 GPU, and Ubuntu 20.04. |
| Software Dependencies | No | The paper mentions 'pre-trained BERT', 'Huggingface' for the BERT model, and 'Adam algorithm' for optimization, but it does not specify software dependencies with version numbers (e.g., Python 3.x, PyTorch x.x, TensorFlow x.x). |
| Experiment Setup | Yes | Specifically, the batch size is set to 8/6 on NYT/Web NLG, and all parameters are optimized by Adam algorithm (Kingma and Ba 2015) with a learning rate of 1e-5. The dimension of the hidden layer de is set to 3 d, the dropout probability in equation (4) is 0.1, the max sequence length is set to 100. |