Abstract:Syntactic processing is fundamental to natural language processing. It provides rich and comprehensive syntax information in sentences that could be potentially beneficial for downstream tasks. Recently, pretrained language models have shown great success in Chinese syntactic processing, which typically involves word segmentation, POS tagging, and dependency parsing. However, the on-going research never ends since performance would be degraded drastically when tested on a highly-discrepant domain. This problem is widely accepted as domain adaptation, where the test domain differs from the training domain in supervised learning. Self-training is one promising solution for it, and straightforward source-to-target adaptation has already shown remarkable effectiveness in previous work. While this strategy ignores the fact that sentences of the target domain sentences may have very different gaps from the source training domain. More specifically, sentences with large gaps might fail by direct self-training adaptation. To this end, we propose fine-grained domain adaptation for Chinese syntactic processing in this work, aiming to model the gaps between the source and the target domains accurately and progressively. The key idea is to divide the target domain into fine-grained subdomains by using a specified domain distance metric, and then perform gradual self-training on the subdomains. We further offer an intuitive theoretical illustration based on the theory of Kumar et al. (2020) approximately. In addition, a novel representation learning framework is proposed to encode fine-grained subdomains effectively, aiming to utilize the above idea fully. Experimental results on benchmark datasets show that our method can achieve significant improvements over a variety of baselines.

Adversarial Domain Adaptation For Chinese Semantic Dependency Graph Parsing

Domain Adaptation for Semantic Parsing.

Semi-supervised Domain Adaptation for Dependency Parsing

Fine-Grained Domain Adaptation for Chinese Syntactic Processing

Dependency-Gated Cascade Biaffine Network for Chinese Semantic Dependency Graph Parsing

Learning Domain Differences Automatically for Dependency Parsing Adaptation.

Semi-supervised Domain Adaptation for Dependency Parsing Via Improved Contextualized Word Representations.

Graph Adaptive Semantic Transfer for Cross-domain Sentiment Classification

Leverage External Knowledge and Self-attention for Chinese Semantic Dependency Graph Parsing

Low-Resource Domain Adaptation for Compositional Task-Oriented Semantic Parsing

Semi-supervised Domain Adaptation for Dependency Parsing with Dynamic Matching Network

Neural Maximum Subgraph Parsing for Cross-Domain Semantic Dependency Analysis.

Tri-training for Dependency Parsing Domain Adaptation

Syntax-Guided Domain Adaptation for Aspect-based Sentiment Analysis

Cross-lingual Dependency Parsing As Domain Adaptation.

Cross-Domain Transfer Learning for Dependency Parsing.

Learning Domain Invariant Word Representations for Parsing Domain Adaptation

A Neural Transition-Based Approach for Semantic Dependency Graph Parsing

Joint Adversarial Domain Adaptation with Structural Graph Alignment

An Empirical Investigation of Domain Adaptation Ability for Chinese Spelling Check Models

Transition-Based Chinese Semantic Dependency Graph Parsing.