基于链接预测的科学研究机会发现方法研究  

A Method for Discovering Scientific Research Opportunities Based on Link Prediction

作　　者：任海英[1] 于立婷[1] 黄鲁成[1,2] Ren Haiying Yu Liting Huang Lucheng(School of Economics and Management, Beijing University of Technology, Beijing 100124 Innovation Center of Capital Society-Building ＆ Social Governanc,Beijing 100124)

机构地区：[1]北京工业大学经济与管理学院,北京100124 [2]首都社会建设与社会管理协同创新中心,北京100124

出　　处：《情报杂志》2016年第10期53-58,36,共7页Journal of Intelligence

基　　金：北京市自然科学基金预探索项目"发明过程和机理的概念地图表示研究"(编号:9153020);北京市教委社会科学计划面上项目"一种基于概念地图的发明过程机理的描述方法"(编号:SM201510005001);国家社会科学基金重大项目"新兴技术未来分析理论方法及产业创新研究"(编号:119ZD140)

摘　　要：[目的/意义]概念的新组合在一定程度上可以带来科学研究上的新突破和新发现,需要探索一套方法来客观有效地挖掘科学知识网络中基于关联关系的研究机会,从而为科研决策提供定量依据。[方法/过程]提出了一种基于链接预测模型的科学研究机会发现方法,以领域中的关键词共现矩阵为数据基础,基于多种评价指标全面评估并选取链接预测算法,最后通过熵权融合指标值,从而预测关键词组合成为有前途研究机会的可能性。[结果/结论]将此方法用于分析和预测自然语言处理领域的研究机会,预测的动态准确率可达50%左右,高于利用单个预测指标的动态准确率,验证了所提方法的可行性和准确性。[ Purpose/Significance] New breakthroughs and discoveries in scientific research can be brought about to a certain extent bynew combinations of existing concepts. It is necessary to explore a method for discovering such research opportunities in the scientificknowledge network objectively and efficiently, and thus provide quantitative basis for scientific research decision making. [ Method/Process] We propose a method for discovering scientific research opportunities based on link prediction, which uses the keywords co-oc-currence matrix as source data and selects link prediction algorithms based on a comprehensive evaluation index. Finally, by using entropycoefficient, the possibility of keyword combinations becoming promising research opportunities is predicted. [ Result/Conclusion ] Thismethod is used to analyze and predict the research opportunities in the field of Natural Language Processing. The dynamic accuracy of theprediction is about 50%, higher than the dynamic accuracy of the single prediction index. Experimental results verify the feasibility and ac-curacy of the method proposed.

关 键 词：科学研究机会 链接预测 共现矩阵 熵权 

分 类 号：G358[文化科学—情报学]