Objective

The Mori Memorial Foundation’s Institute for Urban Strategies, through its Japan Power Cities report, identifies and highlights the special urban characteristics like strengths and attractiveness, of major cities in Japan. However, it does not cover the image or perception held by residents for their cities. Therefore, the purpose of this special research is to offer a potential contribution to the future image branding strategy of each city by clarifying the "city image" held by residents in the form of words and sentences through a questionnaire survey, and by quantitatively analyzing and visualizing them from various perspectives.

1）Resident Questionnaire Survey

Survey period: March, 2022
Respondents: Males and Females above the age of 21, living in one of the JPC cities.
Valid responses: 300 per city
Question: Respondents provide their image of the city they reside in using keywords or short sentences.

Analysis Methodology

• Analysis 1: Top 10 Image Keywords (Frequency Aggregation)

  The top 10 most frequently appearing keywords in each city were extracted from the sentences obtained from the questionnaire responses. These words were obtained after breaking responses down to the smallest unit that had meaning as a word and analyzing them.

• Analysis 2: Relationship between image keywords (co-occurrence network analysis)

  Based on data obtained from the morphological analysis in Analysis 1, keywords that appeared more than 5 times for each city were targeted and a co-occurrence network analysis was undertaken. A network was then drawn with lines connecting those keywords with a strong degree of co-occurrence.

• Analysis 3: Image characteristics for each city (feature word extraction)

  Using the keyword data obtained in Analysis 2, the degree of similarity between texts was measured with the "city of residence" as an external variable. The top 10 keywords that characterize each city were then extracted.

• Analysis 4: Image similarity between cities (co-occurrence network analysis of resident cities and keywords)

  Co-occurrence network analysis was conducted on the keyword data for all target cities used in Analysis 2, with "city of residence" set as an external variable. A network map was then drawn based on the degree of co-occurrence between cities and keywords.

2）Non-Resident Questionnaire Survey

Survey period: March, 2023
Respondents: Males and Females above the age of 21, living in one of the JPC cities.
Valid responses: approx. 900 per city
Question: Respondents provide their image of randomly selected JPC-targeted cities using keywords or short sentences.

Analysis Methodology

• Analysis 1: Image Recognition Percentages

  "Image Recognition Percentages" was conducted for JPC target cities. For the analysis, the "Image Recognition Percentages" was defined as the proportion of responses excluding those with "No Image" answers in each city.

• Analysis 2: Top Image Keywords（Visitors）

  The top most frequently appearing keywords in each city were extracted from the sentences obtained from the questionnaire responses by people who have visited the city before. These words were obtained after breaking responses down to the smallest unit that had meaning as a word and analyzing them.

• Analysis 3: Top Image Keywords（Non-Visitors）

  The top most frequently appearing keywords in each city were extracted from the sentences obtained from the questionnaire responses by people who have never visited the city. These words were obtained after breaking responses down to the smallest unit that had meaning as a word and analyzing them.