38 The Import Structure of LNG from Russia to Japan by Cognitive …
577
38.2 Qualitative Analysis: Cognitive Map
38.2.1 Methodology
The software “KH Coder” (Higuchi 2014, KH Coder) used in this study was developed by Associate Professor Koichi Higuchi of Ritsumeikan University and is free
software that has published all the processing contents since 2001. As of 2018, this
software has been used in more than 2000 research cases, and the fields range from
policy evaluation, media analysis, and verification of educational effects.
Among the functions provided in KH Coder, the following functions were used
in each process in this study.
• Process (1): “extraction word list” “extraction word: hierarchical cluster analysis”
• Process (2): “extraction word: co-occurrence network” “coding: cross tabulation”
Using a database of four major newspaper companies (Nikkei, Yomiuri, Asahi,
and Mainichi) as information sources (Nikkei Telecon, Yomidas Rekishikan, Kikuzo
II Visual, Maisaku), we will extract important terms by text analysis and create a
“cognitive structure map” (Yamamoto and Tani 1979) based on that. It is possible to
extract highly relevant related factors in each period.
38.2.2 Setting Period
In the process of qualitative analysis (1), “cognitive structure diagram” was created
using the functions “extracted word list” and “extracted word: hierarchical cluster
analysis” of KH Coder.
We extracted words that show remarkable appearance throughout the text, and
words that show correlations between events and events of each policy trend such
as “promotion,” “growth,” “interference,” and “deterioration” that are essential for
creating cognitive structure maps. In the process of collecting information, words
such as “Russia” and “Natural gas” are automatically extracted in large numbers, so
we chose not to disturb the analysis.
The function “extraction word: hierarchical cluster analysis” is to output combinations of similar words and phrases of appearance patterns in the form of dendrograms.
Analyze all the articles, grasp the features of the appearance pattern for each frequent
phrase, and automatically connect the words with similar appearance patterns with
closer branches. It is possible to trace the dendrograms back together and group them
together with words connected by close branches as factors that have influenced the
policy. Hereinafter, it is called a related factor.
These factors are then organized into a matrix of this information. At this time,
related factors affecting other related factors are placed in the leftmost column of
the matrix, and related factors affected in the top row of the matrix. After that, based
on the words “promotion” and “prevent” in the sentence, the relationship between
577
38.2 Qualitative Analysis: Cognitive Map
38.2.1 Methodology
The software “KH Coder” (Higuchi 2014, KH Coder) used in this study was developed by Associate Professor Koichi Higuchi of Ritsumeikan University and is free
software that has published all the processing contents since 2001. As of 2018, this
software has been used in more than 2000 research cases, and the fields range from
policy evaluation, media analysis, and verification of educational effects.
Among the functions provided in KH Coder, the following functions were used
in each process in this study.
• Process (1): “extraction word list” “extraction word: hierarchical cluster analysis”
• Process (2): “extraction word: co-occurrence network” “coding: cross tabulation”
Using a database of four major newspaper companies (Nikkei, Yomiuri, Asahi,
and Mainichi) as information sources (Nikkei Telecon, Yomidas Rekishikan, Kikuzo
II Visual, Maisaku), we will extract important terms by text analysis and create a
“cognitive structure map” (Yamamoto and Tani 1979) based on that. It is possible to
extract highly relevant related factors in each period.
38.2.2 Setting Period
In the process of qualitative analysis (1), “cognitive structure diagram” was created
using the functions “extracted word list” and “extracted word: hierarchical cluster
analysis” of KH Coder.
We extracted words that show remarkable appearance throughout the text, and
words that show correlations between events and events of each policy trend such
as “promotion,” “growth,” “interference,” and “deterioration” that are essential for
creating cognitive structure maps. In the process of collecting information, words
such as “Russia” and “Natural gas” are automatically extracted in large numbers, so
we chose not to disturb the analysis.
The function “extraction word: hierarchical cluster analysis” is to output combinations of similar words and phrases of appearance patterns in the form of dendrograms.
Analyze all the articles, grasp the features of the appearance pattern for each frequent
phrase, and automatically connect the words with similar appearance patterns with
closer branches. It is possible to trace the dendrograms back together and group them
together with words connected by close branches as factors that have influenced the
policy. Hereinafter, it is called a related factor.
These factors are then organized into a matrix of this information. At this time,
related factors affecting other related factors are placed in the leftmost column of
the matrix, and related factors affected in the top row of the matrix. After that, based
on the words “promotion” and “prevent” in the sentence, the relationship between