基于语义的疾病相关蛋白质知识抽取  

Semantic output output-based disease-protein knowledge extraction

作　　者：李智恒[1] 杨志豪[1] 林鸿飞[1] 

机构地区：[1]大连理工大学计算机科学与技术学院,辽宁大连116024

出　　处：《山东大学学报（理学版）》2016年第3期104-110,共7页Journal of Shandong University(Natural Science)

基　　金：国家自然科学基金资助项目(61070098;61272373;61340020);新世纪优秀人才支撑计划项目(NCET-13-0084);中央高校基本科研业务费专项资金资助项目(DUT13JB09;DUT14YQ213)

摘　　要：随着人类基因组学研究和高通量技术的发展,涉及蛋白质知识以及相关疾病、药物的医学文献呈指数增长。利用文本挖掘技术从大量的生物医学文本中发现和抽取有价值的、新颖的蛋白质知识已经成为可能。基于SemRep得到的特定疾病的M EDLINE文献的语义输出,通过显著信息提取算法对该语义输出进行打分排序,抽取得到与特定疾病相关的蛋白质以及蛋白质和药物之间的联系。之后与KEGG数据库中列出的该疾病相关的蛋白质、基因进行比较。实验结果对理解疾病的病因、蛋白质功能预测以及药物辅助设计都有重要的研究意义。With the rapid development of genomics and high-throughput technologies,large amount of biomedical literatures about genes and proteins appear. M eanwhile,the use of text mining technology discovery and excavation of new,valuable knowledge of protein from the mass of medical texts has become possible. This paper presents a system which extracts the relations between proteins and certain diseases from biomedical literature based on semantic output generated by SemRep,and then extracts novel,valuable protein knowledge. The system summarizes the salient relations by the salient extraction algorithm using the specific subject M EDLINE corpus. Subsequently,the results extracted by the system are compared with data from KEGG database. Implementation of the system has important significance for understanding the major causes of many diseases,protein function prediction and drug-aided design.

关 键 词：语义关系 信息抽取 SemRep KEGG 

分 类 号：R3411[医药卫生—基础医学] TP391.1[自动化与计算机技术—计算机应用技术]