Abstract:Due to the polysemy and complexity of the Chinese language, Chinese machine reading comprehension has always been a challenging task. To improve the semantic understanding and robustness of Chinese machine reading comprehension models, we propose a model that utilizes adversarial training algorithms and Permuted Language Model (PERT). Firstly, we employ the PERT pre-training model to embed paragraphs and questions into vector space to obtain corresponding sequential representations. Secondly, we use a multi-head self-attention mechanism to extract key textual information from the sequence and employ a Bi-GRU network to semantically fuse the output feature vectors, aiming to learn deep semantic representations in the text. Finally, we introduce perturbations into the model training process. We achieve this by utilizing adversarial training algorithms such as Fast Gradient Method (FGM) and Projected Gradient Descent (PGD). These algorithms generate adversarial samples to enhance the model's robustness and stability when facing diverse inputs. We conducted comparative experiments on the publicly available Chinese reading comprehension datasets CMRC2018 and DRCD. The experimental results show that our proposed model has achieved significant improvements in both EM and F1-Score compared to the baseline model. To validate the model's generalization and robustness, we utilized ChatGPT to construct a scientific dataset that includes a large number of domain-specific terms, sentences with mixed Chinese and English, and complex comprehension tasks. Our model also performed remarkably well on the self-built dataset. In conclusion, the proposed model not only effectively enhances the understanding of semantic information in Chinese text but also demonstrates a certain level of generalization capability.

A Hybrid Model for Chinese Confusable Words Distinguishing in Proofreading

A Hybrid Model for Chinese Spelling Check

A Hybrid Chinese Language Model based on a Combination of Ontology with Statistical Method

A Hybrid Method for Entity Hyponymy Acquisition in Chinese Complex Sentences.

Human-Computer Interactive Chinese Word Segmentation: an Adaptive Dirichlet Process Mixture Model Approach.

A Hybrid Approach to Chinese-English Machine Translation

A Hybrid Approach of Text Segmentation Based on Sensitive Word Concept for NLP

A Unified Model for Joint Chinese Word Segmentation and POS Tagging with Heterogeneous Annotation Corpora.

Exploiting Word Semantics to Enrich Character Representations of Chinese Pre-trained Models

A Hybrid Post-Processing System For Offline Handwritten Chinese Character Recognition Based On A Statistical Language Model

An Adversarial Multi-Task Learning Method for Chinese Text Correction with Semantic Detection

Combining permuted language model and adversarial training for Chinese machine reading comprehension

Chinese Sentence Similarity Computational Model Based on Multi-Features Combination

Chinese Lexical Analysis Based On Hybrid Mmsm Model

A Hybrid Model for Grammatical Error Correction.

Mitigating Idiom Inconsistency: A Multi-Semantic Contrastive Learning Method for Chinese Idiom Reading Comprehension

A Unified and Discriminative Soft Syntactic Constraint Model for Hierarchical Phrase-based Translation.

Do We Need Language-Specific Fact-Checking Models? The Case of Chinese

A Model for Chinese Sentence Similarity Computing

A Hybrid Language Model Based on Statistics and Linguistic Rules

Exploring Hybrid Character-Words Representational Unit in Classical-to-Modern Chinese Machine Translation