MINING DOMAIN KNOWLEDGE FROM SCIENTIFIC DOCUMENT

摘要

Scientific document is a principal instrument for the cumulation and coordination of domain knowledge. Due to the recent availability of internet-based searching services and scientific databases, we have easies access to scientific document. We propose a method based on bibliometric analysis to mine domain knowledge from scientific document. It includes term extraction, seed term selection, linear regression analysis on seed term, seed term clustering, linkage analysis of term. In order to help researchers to identify the technology trend and research direction in the future, we define several indicators including contribution of seed tenn, promising rate of seed term, borrowing rate of term. Then, we apply our method to the domain of nanotechnology. In contrast to the expert-based approach, the method has more objectivity. It does not have the preconceived limitations, constraints, biases, and personal and organizational agendas of the experts. It can help identify promising science and technology directions and provide new opportunities of knowledge discovery for researchers. The method does not obviate the need for detailed investigation of the literature or interactions with the experts in order to make a substantial contribution to the understanding and the advancement of some domains. It helps to allow these detailed efforts to be executed more efficiently.