摘要
随着人工智能技术的不断发展,深度学习方法在数学单词问题智能求解方面得到了广泛应用。基于图到树的编解码网络表现出良好的性能,但外部知识缺乏使得模型求解准确率提升受到限制,构建一种基于几何知识库的检测模块,将其融入编解码器网络模型,提高了模型的预测能力。通过在中文数学问题数据集GeometryQA上进行验证,模型表现出更高的准确率,具有一定的优越性。
With the continuous development of artificial intelligence technology,deep learning methods have been widely used in the intelligent solution of math word problem.Graph-to-tree based encoder-decoder networks show good performance,but the lack of external knowledge limits the improvement of model solving accuracy.A detection module based on a geometric knowledge base was constructed and incorporated into the encoder-decoder net-work model to improve the predictive capability of the model.Through validation on the Chinese mathematical problem dataset GeometryQA,the model exhibits higher accuracy and has certain superiority.
作者
魏亚琴
刘斌
张倩
崔学英
谢秀峰
WEI Ya-qin;LIU Bin;ZHANG Qian;CUI Xue-ying;XIE Xiu-feng(School of Applied Science,Taiyuan University of Science and Technology,Taiyuan 030024,China)
出处
《太原科技大学学报》
2024年第5期507-513,共7页
Journal of Taiyuan University of Science and Technology
基金
国家自然科学基金(11701406,11901134,12061091,11972019)
山西省基础研究计划项目(202103021224274,201901D111261)
山西省社会经济统计科研课题(KY[2022]73)
山西省省筹资金资助回国留学人员科研项目(2022-163)。
关键词
自然语言处理
数学单词问题
几何知识库
检测模块
natural language processing
mathematical word problem
geometric knowledge base
detection network
