ANALYZING INTERNATIONAL COLLABORATION AND IDENTIFYING CORE TOPICS FOR THE "INTERNET OF THINGS" BASED ON NETWORK ANALYSIS AND TOPIC MODELING

Abstract

The study of the "Internet of Things (IoT)" has been consistently evolving and rapidly expanding in the last decade. This research aims to examine the international collaboration among countries, identify major research areas and core topics in IoT-related studies, and exploit their relationships and dynamic changes over time. First, we construct a co-work network among countries and compare centrality values to identify which countries are actively involved in international collaboration. To identify major research areas and core topics, we employ a two-layer approach: main research areas are determined by word co-occurrence analysis based on titles and author-provided keywords; core topics are found using a latent Dirichlet allocation (LDA), which is a well-known topic model that automatically discovers latent topics pervading a large collection of documents. Topic trends are also analyzed by modeling the topic proportions over time as a simple linear regression to determine hot and cold topics. A topic network integrating the discovered main research areas and core topics is also constructed to identify meaningful relationships among them. Based on 8,582 research papers published from 2003 to 2015, it was found that China and the United States are the two leading countries in terms of publication volume. The United States is also actively involved in international collaboration. The role of China in international collaboration is not as significant as in publication volume. Eight main research areas were identified by title and keyword network analysis, and 20 core topics were discovered by LDA. Among these 20 topics, six of them are gradually emerging while four of them are declining.