基于文档集的生物信息挖掘模型研究  

Research on biological information mining model based on document set

作　　者：孙红敏[1] 姜楠楠[1] 李想[1] SUN Hongmin;JIANG Nannan;LI Xiang(School of Electrical and Information, Northeast Agricultural University, Harbin 150030, China)

机构地区：[1]东北农业大学电信与信息学院,哈尔滨150030

出　　处：《计算机工程与应用》2016年第24期102-106,188,共6页Computer Engineering and Applications

摘　　要：针对生物医学文献的数量急剧增长,人工从文献中获取所需要的信息已不能适应生物医学文献数量迅速生长的需要。利用Stanford Parser等开源工具,采用自然语言处理技术、统计学等多种方法,提出了一种新型的生物信息挖掘模型,并对其关键技术进行分析。该模型在对全文文本SBQTL(Soybean Quantitative Trait Loci)测试中父母本信息提取的准确率和召回率分别为93.0%和78.4%;在对Pub Med测试中,准确率和召回率分别为94.3%和80.0%。解决了生物医学研究者从海量文献中更有效、快速地找到所需信息的问题,以便生物学家发现隐藏的生物医学知识并验证得到新的科学发现,从而使人们对生物医学现象的认识得到了提高。As the quantity of literature increases dramatically, to get the information manully can’t adapt to the speed of added literature. This paper proposes a new model of biological data mining, utilizing some tools of open source such as Stanford Parser, using some approaches such as natural language processing and statistics. It also analyzes its crucial technique. During the process to test the SBQTL(Soybean Quantitative Trait Loci)using this model, the precision and recall rate are 93.0% and 78.4% respectively. During the process to test the PubMed, the precision and recall rate are 94.3% and 80.0% respectively. So the problem that the researchers who are engaged in biomedicine can find the information they need from large quantity of literature quickly and efficiently is solved, and biologists can find closet information in biomedicine and verificate the newest science discovery. Thus, people can better understand the phenomenon of biomedicine.

关 键 词：文本挖掘 STANFORD PARSER 文本预处理 依存关系 信息抽取 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]