Empirical Analysis of Unlabeled Entity Problem in Named Entity Recognition

Authors: Yangming Li, lemao liu, Shuming Shi

ICLR 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Through empirical studies performed on synthetic datasets, we find two causes of performance degradation.
Researcher Affiliation Industry Yangming Li, Lemao Liu, & Shuming Shi Tencent AI Lab {newmanli,redmondliu,shumingshi}@tencent.com
Pseudocode No The paper describes methods in prose and with diagrams (Figure 3) but does not include any explicit pseudocode or algorithm blocks.
Open Source Code Yes 1Our source code is available at https://github.com/LeePleased/NegSampling-NER.
Open Datasets Yes The material datasets are Co NLL-2003 (Sang & De Meulder, 2003) and Onto Notes 5.0 (Pradhan et al., 2013).
Dataset Splits Yes The data contains 2400 sentences tagged by human annotators and are divided into three parts: 1200 for training, 400 for dev, and 800 for testing.
Hardware Specification No The paper does not provide specific details about the hardware (e.g., GPU/CPU models, memory) used for running the experiments.
Software Dependencies No The paper mentions using Adam for optimization but does not provide specific version numbers for software dependencies or libraries (e.g., Python, PyTorch, TensorFlow versions).
Experiment Setup Yes L2 regularization and dropout ratio are respectively set as 1 10 5 and 0.4 for reducing overfit. The dimension of scoring layers is 256. Ratio λ is set as 0.35. When the sentence encoder is LSTM, we set the hidden dimension as 512 and use pretrained word embeddings (Pennington et al., 2014; Song et al., 2018) to initialize word representations. We utilize Adam (Kingma & Ba, 2014) as the optimization algorithm and adopt the suggested hyper-parameters.