基于强化学习的古今汉语句子对齐研究  

Sentence Alignment of Ancient and Modern Chinese Based on Reinforcement Learning

作　　者：喻快 邵艳秋 李炜[1] YU Kuai;SHAO Yanqiu;LI Wei(National Language Resources Monitoring and Research for Print Media Language Center,Beijing Language and Culture University,Beijing 100083,China)

机构地区：[1]北京语言大学国家语言资源检测与研究平面媒体中心,北京100083

出　　处：《中文信息学报》2024年第12期30-38,63,共10页Journal of Chinese Information Processing

基　　金：国家自然科学基金(61872402);教育部人文社科规划基金(17YJAZH068);北京语言大学校级项目(中央高校基本科研业务费专项基金)(21YBB19,18ZDJ03);模式识别国家重点实验室开放课题基金。

摘　　要：基于深度学习的有监督机器翻译取得了良好的效果,但训练需要大量高质量的对齐语料。对于中文古今翻译场景,高质量的平行语料相对匮乏,这使得语料对齐在该领域具有重要的研究价值和必要性。在传统双语平行语料的句子对齐研究中,传统方法根据双语文本中的长度、词汇、共现文字等特征信息建立综合评判标准来衡量两个句对的相似度。此类方法对句子语义匹配的能力有限,并且在多对多的对齐模式上表现不佳。该文利用具有强大语义能力的预训练语言模型,并基于动态规划算法的强化学习训练目标来整合段落全局信息,进行无监督训练。实验结果证明,使用该方法训练得到的模型性能优于此前获得最好表现的基线模型,特别是在多对多对齐模式下,性能提升显著。Supervised machine translation based on deep learning has achieved good results,which it requires a large amount of high-quality aligned parallel corpora for training.For the Chinese historical and modern translation scenario,the relative scarcity of high-quality parallel corpora highlights the significant research value and necessity of text alignment in this field.In traditional research on bilingual sentence alignment,conventional methods use features such as length,vocabulary,and co-occurring characters in bilingual texts to establish a comprehensive evaluation criterion for measuring the similarity between two sentence pairs.Such methods have limited ability to match sentence semantics and perform poorly in the many-to-many alignment mode.This paper proposes using pre-trained language models with powerful semantic capabilities and reinforcement learning training objectives based on dynamic programming algorithms to integrate paragraph-level global.Experimental results demonstrate that the model trained using the method proposed in this paper outperforms the previously best-performing baseline models,with particularly significant improvements in handling many-to-many alignment patterns.

关 键 词：双语对齐 预训练语言模型 强化学习 动态规划 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]