The tourism industry is attracting attention as a powerful means of regional revitalization, and the needs and impressions of actual tourists are very important in developing the tourism industry. In this study, using review posted on Japanese travel information websites, features and impressions of tourist spots held by actual tourists were quantitatively identified by text mining. The target regions for quantifying the features and impressions of tourist attractions held by actual tourists were Hakui, Kaga, and Wajima, which are cities in Ishikawa Prefecture that differ in the analysis of regional features. In this study, reviews posted to tourist spots in each region were collected by scraping using Jalan.net, one of the leading travel reservation sites in the field of domestic travel in Japan. Text mining of the collected word-of-mouth data was used to understand the features and impressions of tourist attractions in each region held by actual tourists. In order to conduct text mining, morphological analysis was conducted on all word-of-mouth data using KHCoder. Using this review data, the features of the actual tourist attractions in the three regions were calculated by selecting 15 nouns from the extracted nouns in each region in descending order of the number of extracted nouns. The impressions of tourist attractions in each region were obtained by selecting three representative nouns from the extracted nouns and calculating the adjectives and adjectival verbs that are strongly related to these nouns. The selected nouns reflect the features of the tourist spots in the region, and the impressions are the general impressions of the selected nouns. This suggests that it is possible to quantify the regional features and impressions of tourist spots held by actual tourists from review.