基于知识网络的科学研究机会发现的机理和应用  

Mechanisms and Applications of Scientific Research Opportunities Discovery Based on Knowledge Network

作　　者：任海英[1] 赵育慧 于立婷[1] 

机构地区：[1]北京工业大学经济与管理学院北京现代制造业研究基地,北京100124

出　　处：《情报理论与实践》2018年第11期89-95,共7页Information Studies:Theory & Application

基　　金：国家自然科学基金面上项目"基于多源异构数据的新兴技术形成机理研究"的成果;项目编号:71673018

摘　　要：[目的/意义]识别潜在的科学研究机会是科研人员进行研究的前提,探讨科学研究机会发现的机理和方法有助于提高科研人员的研究效率。[方法/过程]综合相关文献知识发现和技术机会发现的理论和方法,对科学研究机会进行概念界定和形态表示,并以知识网络的形式阐述了4种发现机理,最后基于链路预测的方法,对单词级别的知识网络构建科学研究机会发现模型。[结果/结论]将该模型应用于自然语言处理领域,预测动态准确率达75. 1%,预测结果表明该模型能够有效进行科学研究机会发现。在理论上填补科学研究机会发现机理研究的空白,方法上将预测的知识单元扩展到单词级别,从而从新的角度发现具有创造性的科学研究机会。[ Purpose/significance ] Identifying potential scientific research opportunities is a premise for conducting research, and exploring mechanisms of and methods for discovering scientific research opportunities can help increase the research efficiency of scientific researchers. [ Method/process ] Firstly, the theories and methods of literature-based discovery and technology opportunity discovery are integrated in order to define and classify scientific research opportunities conceptually. Then, four mechanisms of dis- covery are explicated in the form of knowledge networks. Finally, link prediction method is used to construct models for scientific research opportunities discovery based on word-level knowledge networks. [ Result/conclusion ~ The model is applied in the field of natural language processing with dynamic accuracy rate of 75.1%. The prediction results show that the model can be effectively used for discovering scientific research opportunities. This work fills a theoretical gap in the mechanism of scientific research oppor- tunities discovery, and the proposed method extends the prediction of knowledge unit to word level, thus creative scientific research opportunities can be discovered from a new perspective.

关 键 词：科学研究机会 知识网络 链路预测 自然语言处理 

分 类 号：G301[文化科学] G353.1