Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].
Domain-Adapted Dependency Parsing for Cross-Domain Named Entity Recognition
Authors: Chenxiao Dou, Xianghui Sun, Yaoshu Wang, Yunjie Ji, Baochang Ma, Xiangang Li
AAAI 2023 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Finally, through extensive experiments, we show our proposed method can not only effectively take advantage of word-dependency knowledge, but also significantly outperform other Multi-Task Learning methods on cross-domain NER. |
| Researcher Affiliation | Collaboration | Chenxiao Dou1, Xianghui Sun2, Yaoshu Wang *3, Yunjie Ji2, Baochang Ma2, Xiangang Li2 1Nanhu Academy of Electronics and Information Technology 2Beike 3Shenzhen Institute of Computing Sciences, Shenzhen University EMAIL, EMAIL, EMAIL |
| Pseudocode | No | The paper does not contain structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | Our code is open-source and available at https://github. com/xianghuisun/DADP. |
| Open Datasets | Yes | To evaluate the effectiveness of the proposed method, we conduct experiments on four English NER datasets, including Co NLL03 1, WNUT17 2, Mit Rest 3 and NCBI 4. The four datasets come from four different domains, which are listed in Table 1. In addition, as DP task is taken as the auxiliary task in the proposed method, we adopt Onto Notes 5.0 5 as our DP source dataset, converted to the Stanford dependency-tree format by using Stanford Core NLP (Manning et al. 2014). Detailed statistics of the datasets are listed in Table 1. |
| Dataset Splits | No | Table 1 shows the statistics for 'Train' and 'Test' splits (e.g., 'Onto Notes #sentence 59924 8262'), but does not explicitly provide details for a validation split. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., exact GPU/CPU models, processor types, or memory amounts) used for running its experiments. |
| Software Dependencies | No | The paper mentions software like BERT, Bi LSTM, Adam W, and spaCy, but does not provide specific version numbers for these or any other ancillary software components. |
| Experiment Setup | Yes | Hyperparameters. We set the threshold of the maximum epoch as 100 for every model training. To our proposed model, the adopted Bi LSTM module is incorporated with two 768-dimension LSTM layers. Each representation layer after Bi LSTM is introduced with 128 dimensions. For the two biaffine classifiers, the parameters are configured as described in the previous section.With all the datasets, we use the batch size as 16 and the input maximum length as 256. In the training process, Adam W is taken as our optimizer with the learning rate 2e-5. |