Academic Research Trend Analysis Using Social Network Analysis and Text Mining: Case of Cloud Computing Papers

摘要

In the paper, we propose an academic research trend analysis technique based on social network analysis and text mining. The technique provides a framework to understand dynamic changes on research trends based on analysis of research papers' bibliographic information. To track dynamic research trends, term networks are constructed annually. To build term networks, abstracts of papers are manipulated using text mining techniques. Even though abstracts in papers provide rich sets of information of papers, it is difficult to extract meaningful terms which are related to cloud computing. To define a set of useful terms, a set of whole keywords in articles are used. Based on eo-occurrence relationships among terms, term networks are generated. To visualize research trend, we propose research trend map in which terms are pointed in a two dimensional space. The X-axis is normalized term frequency and Y -axis is normalized degree centralities in a research trend map. Annual changes of normalized term frequencies of a term show the changes of popularity of the term. Also, annual changes of normalized degree centralities of a term show the changes of centrality of the term in a specific research domain. To show the usefulness of the proposed approach, the approach has been applied to analyze research trend of cloud computing. Paper articles which include "cloud computing" term from 1994 to 2013