基于PERT的中医药知识抽取式问答模型研究

Research on Extractive Q&A Model for Traditional Chinese Medicine Knowledge Based on PERT

中医药领域积累了丰富的知识与经验,但如何从这些海量、深奥的中医资料中准确提取中医药知识,一直是医学领域的挑战。为了提供高效准确的中医药知识抽取方法,提出了一种基于PERT模型的中医药知识抽取式问答模型。该方法依托中医药领域的专业知识与增强数据集,结合PERT模型,使用乱序语言预训练任务,实现了一个具有较强中医药知识阅读理解能力的问答模型。实验结果表明,该模型在中医药知识数据集上的问答性能优于其他相关模型,当给出中医药知识文本和问题时,能较为精确地理解并给出对应答案。

The field of Traditional Chinese Medicine(TCM)has accumulated a wealth of knowledge and experience,but how to accurately extract TCM knowledge from these massive and profound TCM materials has always been a challenge in the medical field.In order to provide an efficient and accurate method for extracting traditional Chinese medicine knowledge,a traditional Chinese medicine knowledge extractive Q&A Model based on the PERT model is proposed.The method relies on professional knowledge and enhanced datasets in the field of TCM,and uses the PERT model and disordered language to pretraining task,and a Q&A model with strong reading comprehension of TCM knowledge is realized.Experiment results show that the Q&A performance of this model on the TCM knowledge dataset outperforms that of other related models,and it can understand and give the corresponding answers more accurately when the TCM knowledge text and questions are given.