Recently, people are active in sharing their experiences with visiting trendy places on Social Network Service (SNS). Such places are expressed as ‘hot place’, and the unique atmosphere of hot places including restaurants have become a major reason to visit. This study used semantic network analysis to examine the images of hot places mentioned in social media focusing on restaurants by limiting the spatial scope to Seoul Metropolitan City. This study used the Textom for Text Mining using data from blogs of NAVER and DAUM, news articles of Google, Facebook, and Twitter. To examine the seasonal effect, data from two seasons (summer and winter) were collected and analyzed accordingly. This study verified the images of hot places by conducting the centrality analysis using the UCINET, and classified similar keywords into clusters through the structural equivalence analysis. The results showed that Seongsu-dong, Hongdae, Ikseon-dong, and Eulji-ro were mentioned the most as hot places of Seoul region. The keywords such as ‘cafe’ and ‘date’ were mostly searched. The images of places were verified as follows: Seongsu-dong as industrial interior design and remodeling factories, Hongdae as a place with great clubs and cost effectiveness, Ikseon-dong as new-tro using Hanok, and Eulji-ro as a place coexisting with the past and present. It has been found that the image of a region is variously recognized, and the image of a restaurant food and the image of a place do not necessarily match. The findings could serve as useful information for restaurant start-ups and local marketing.