基于深度学习的PPI关系抽取方法研究进展  

RESEARCH PROGRESS OF PROTEIN-PROTEIN INTERACTION RELATIONSHIP EXTRACTION METHODS BASED ON DEEP LEARNING

作　　者：唐詹 王龙鹤[1] 郭旭超 周晗 刁磊 李林[1] Tang Zhan;Wang Longhe;Guo Xuchao;Zhou Han;Diao LeiLi Lin(College of Information and Electrical Engineering,China Agricultural University,Beijing 100085,China)

机构地区：[1]中国农业大学信息与电气工程学院,北京100085

出　　处：《计算机应用与软件》2023年第5期1-9,共9页Computer Applications and Software

基　　金：国家重点研发计划项目(2016YFD0300710)。

摘　　要：蛋白质间相互作用(protein-protein interaction,PPI)在几乎全部的生命活动中发挥着极其重要的作用,如何准确、高效地自动化抽取出科学文献中的PPI关系有着十分重要的研究意义。深度学习作为机器学习的一个重要分支,由于其强大的学习能力和特征提取能力,在PPI关系抽取领域得到了广大研究者的关注。对现有基于深度学习的PPI关系抽取方法进行阐述,对这些方法的实验数据及其构成组件进行对比分析,总结这些方法存在的问题,展望可以改进的研究方向。Protein-protein interaction(PPI)plays an extremely important role in almost all life activities.How to automatically extract the PPI relationships in scientific literature accurately and efficiently has a very important research significance.As an important branch of machine learning,deep learning methods has drawn the attention of researchers at large in the field of PPI relationship extraction,due to its powerful learning ability and feature extraction ability.Existing PPI relationship extraction methods based on deep learning were described,and the experimental data and components of each method were compared and analyzed.The problems which these methods exist were summarized,and the directions that can be improved were forecasted.

关 键 词：蛋白质间相互作用 关系抽取 深度学习 自然语言处理 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]