Cities attract people by holding many events such as seasonal product sales, festivals, exhibitions, and other trendy events. Information on the various events spreads on the internet by the media. This suggests that media articles can be real-time outputs of the city's transition associated with events. However, the accumulation of information in the media, which is updated daily, has not been sufficiently managed, and the city has not been evaluated from the perspective of events held there. This study aims to examine the transition or flow of events extracted from media articles by analyzing an online database of event announcements.

In this study, we analyze articles published on FASHION PRESS. We construct a spatio-temporal directed graph by using the data of the coordinates and dates of events in Tokyo. The node of the graph represents an event. The weight of each edge is given by considering the Euclidean distance, the difference in dates, and the similarity of the article texts between two events. We constructed the Minimum Spanning Tree (MST) of the graph to visualize the transition of events in a city. We change the way to assign the weights to each edge according to the difference in the place and date of the events parametrically and compare the output graphs. By analyzing these MSTs, we evaluate the image of the city in terms of changes in the content of events, such as the occurrence of trends and seasonal changes.

By comparing the urban image obtained from media articles with the actual city and its concrete urban planning, we can extend the image of the city obtained through conventional fieldwork to that incorporating information space. Such an extension will allow us to propose a new perspective of urban planning considering digital urban space.