Abstract:Semantic role labeling functions to convey the meaning of a sentence through forming a predicate-argument structure directed at the specific predicate. In recent years, end-to-end semantic role labeling methods associated with the deep neural network have received significant attention in the field of computational linguistics. Moreover, end-to-end semantic role labeling methods have demonstrated a beneficial capacity to reduce the incompleteness caused by handcrafted features, which is an observed short-coming of traditional Chinese role labeling methods. However, the critical focus of sentences are frequently lost as a result of existing semantic role labeling structures attributing equal importance to every single word, instead of the overall concept denoted by particular terms. Hence, the performance and function ability of deep neural network models is reduced. In this paper, we introduce a specific attention mechanism based on the established predicate. This mechanism would automatically calculate the weighted contributions of each word, and the corresponding Part-of-Speech, in order to accurately represent the general fundamental ideas of the sentence. In addition, we extended the Bidirectional LSTM using two different semantic role constraint methods, to effectively utilize the dependency and constraint relationships among different semantic role tags, hereby further improving the performance of the whole neural Chinese semantic role labeling model. Experimental results demonstrate the efficacy of our proposed model through providing a baseline that allows for meaningful comparisons, inferring that both weighted contributions of the predicate, and semantic role constraints can help significantly refine the overall model function.

Position-aware Self-Attention Based Neural Sequence Labeling

A Relative Position Attention Network for Aspect-Based Sentiment Analysis

Jointly Trained Sequential Labeling and Classification by Sparse Attention Neural Networks.

Citation Metadata Extraction Via Deep Neural Network-based Segment Sequence Labeling

ASRNN: A recurrent neural network with an attention model for sequence labeling

Label Attention Network for Structured Prediction

Spatial Context-Aware Object-Attentional Network for Multi-Label Image Classification

Neural Attention Models for Sequence Classification: Analysis and Application to Key Term Extraction and Dialogue Act Detection

Label Attention Network for sequential multi-label classification: you were looking at a wrong self-attention

End-to-end Sequence Labeling via Bi-directional LSTM-CNNs-CRF

Adaptive Self-training for Few-shot Neural Sequence Labeling

Neural Latent Dependency Model for Sequence Labeling

Rethinking Self-Attention: Towards Interpretability in Neural Parsing

Joint Segmentation and Tagging with Coupled Sequences Labeling

Predicate-attention neural model for Chinese semantic role labeling

Empower Sequence Labeling with Task-Aware Neural Language Model

Improving Self-Attention Networks with Sequential Relations

A Label-Specific Attention-Based Network with Regularized Loss for Multi-label Classification

Reinforced Self-Attention Network: a Hybrid of Hard and Soft Attention for Sequence Modeling

Combining Discrete and Neural Features for Sequence Labeling

A Survey on Recent Advances in Sequence Labeling from Deep Learning Models