疾病-病症和病症-治疗物质的关系抽取研究

发布时间：2018-02-27 15:14

本文关键词： 信息抽取半监督学习 Tri-training 集成学习　出处：《计算机工程与应用》2017年10期 　论文类型：期刊论文

【摘要】：随着生物医学文献的快速增长,在海量的生物医学文献中存在大量有关疾病、病症和治疗物质的信息,这些信息对疾病的治疗和药物的研制有着重要的意义。针对疾病与治疗物质之间的信息抽取,重点训练两个模型,即疾病与病症模型和病症与治疗物质模型。疾病与病症模型判断一种疾病是否会存在或者导致一种生理现象的产生;病症与治疗物质模型判断一种物质是否改变人的生理现象或者生理过程。使用半监督学习的Tri-training的方法,利用大量未标注数据辅助少量有标注数据进行训练提高分类性能。实验结果表明,Tri-training方法中利用未标注数据有助于提高实验结果;且在训练过程中使用集成学习的思想将三个分类器器集成在一起,提高了学习性能。
[Abstract]:With the rapid growth of biomedical literature, there is a large amount of information about diseases, diseases and therapeutic substances in the vast amount of biomedical literature. This information is of great significance to the treatment of diseases and the development of drugs. Disease and disease model and disease and treatment material model. Disease and disease model to determine whether a disease will exist or lead to a physiological phenomenon; Disease and Therapeutic substances Model determines whether a substance changes a person's physiological phenomena or processes. A semi-supervised learning method called Tri-training, A large amount of unlabeled data is used to train a small amount of labeled data to improve the classification performance. The experimental results show that the use of unlabeled data in the Tri-training method is helpful to improve the experimental results. The three classifiers are integrated in the training process using the idea of integrated learning, which improves the learning performance.
【作者单位】：大连理工大学计算机科学与技术学院;
【基金】：国家自然科学基金(No.61070098,No.61272373,No.61340020) 新世纪优秀人才支撑计划(No.NCET-13-0084) 中央高校基本科研业务费专项资金(No.DUT13JB09,No.DUT14YQ213)
【分类号】：TP181;TP391.1

【相似文献】