Proposal / Research

Research Paper

Visualizing the annual transition of ocean policy in Japan using text mining

A research group led by Dr. Zhu Mengyao, a research fellow of Ocean Policy Research Institute of the Sasakawa Peace Foundation (President: Dr. Atsushi Sunami), has used unsupervised machine learning to classify years of text data in multiple ocean-related white papers to comprehensively visualize annual trends in ocean policy in Japan, and to identify similarities and differences in the concerns of the ocean-related organizations that publish the white papers. They have also identified similarities and differences in the concerns of ocean-related organizations, and verified the effectiveness of text mining methods for examining ocean policy.

This research has been published from Marine Policy on 10 July 2023 (JST).
PDF of full issue is available from ScienceDirect


This study visualized the transition of ocean policy in Japan using text mining technique applied to the text data of three annual reports: the Fishery white paper from 2007 to 2020, Environmental white paper from 2008 to 2020, and Ocean white paper from 2004 to 2020. The results of the analysis were compared with the expert interviews on each subject. Based on latent Dirichlet allocation (LDA) topic model analysis, significant differences in topics were observed among these white papers, while they all clearly responded to the Great East Japan Earthquake occurred in 2011 with their unique topics. Major transitions in topics occurred several times over the years for all white papers. For the Fishery white paper and Environmental white paper issued by the government agencies, which were mainly caused by the revisions of relevant laws; while for the Ocean white paper issued by non-government organization,which were closely related to the development of worldwide ocean initiatives. The expert interviews revealed that the experts’ views on the topics and the transitions of policy focuses were generally consistent with the unsupervised analysis results. Automated visualization of policy transition in each organization accelerates the extraction of their directions and roles in policymaking which will help provide scientific evidence for identifying future opportunities for inter-organizational collaboration.

Page Top