Abstract: Fast discovery and accurate grasp of research frontier is not only the focus of scientific research, also is the need for strategy decisions of top managers and governments. In order to detect research frontier of specific domain accurately and comprehensively using objective methods, firstly, this paper defines the definition of research frontier and analyzes its characteristics. Then, we suggest innovativeness and interdisciplinarity-based detection model of research frontier. The suggested model consists of "three-process" that are identifying innovativeness and interdisciplinarity and refining frontier terms. In this processes, they are analyzed using literature citations, co-occurrence analysis and text mining methods. Finally, we use suggested model to detect research frontier of smart materials. It explains using process and results of suggested model. Experimental result shows a higher degree match of forecasting frontier terms using suggested model and experts' judgements of the field of smart materials.

摘要： 快速发现并科学准确地把握研究前沿不仅是科学研究人员关注的焦点，更是高层管理者乃至各国政府战略决策的需要。为了从客观角度全面准确地探测特定领域的研究前沿，论文首先界定研究前沿的内涵并分析其特征，进而提出从创新性和学科交叉性探测研究前沿的模型，此模型由“三阶段”构成，即创新性特征识别阶段、学科交叉性特征识别阶段以及前沿主题提炼阶段，处理过程分别使用文献引文分析、共现分析和文本挖掘方法，最后以智能材料领域为例，说明模型（方法）的使用过程和效果。实验结果表明由此模型探测的前沿主题与智能材料领域专家的判断相符程度较高。