Uncovering inconspicuous places using social media check-ins and street view images. (May 2020)

• Record Type: Journal Article
• Authors: Zhang, Fan; Zu, Jinyan; Hu, Mingyuan; Zhu, Di; Kang, Yuhao; Gao, Song; Zhang, Yi; Huang, Zhou
• Is Part Of: Computers, environment and urban systems. Volume 81(2020)
• Publication Date: 2020-05
• Subjects: Place semantics -- Social sensing -- Residents and visitors -- Social media check-ins -- Street view images; City planning -- Data processing -- Periodicals; Regional planning -- Data processing -- Periodicals; 303.4834
• DOI: 10.1016/j.compenvurbsys.2020.101478
• Languages: English
• ISSNs: 0198-9715

Abstract

Abstract: There is a Chinese proverb, "if your wine tastes really good, you do not need to worry about the location of your bar (酒香不怕巷子深)", which implies that the popular places for local residents are sometimes hidden behind an unassuming door or on unexpected streets. Discovering these unassuming places (e.g. restaurants) of a city will benefit the understanding of local culture and help to build livable neighborhoods. Previous work has been limited by the lack of appropriate data sources and efficient tools to evaluate the popularity, ambiance and physical surroundings of places in large-scale urban areas. In addition, how to characterize places with respect to different groups of people remains unclear. In this work, we propose a data-driven approach using social media check-ins and street-level images to compare the different activity patterns of visitors and locals, and uncover inconspicuous but interesting places for them in a city. We use check-in records as a proxy of the popularity of a particular type of place, and differentiate visitors and locals based on their travel and social media behaviors. In addition, we employ street-level images to represent the physical environments of places. As a result, we discovered a number of inconspicuous yet popular restaurants in Beijing. These restaurants are located mostly in deep alleys of Old Beijing neighborhoods, where the physical environments are not particularly appealing; however, these places are frequently visited by locals for social engagements. We also discovered beautiful but unpopular outdoor places in Beijing. These places are potential recreational areas for all groups of people and could be improved regarding urban design and planning to make these public infrastructures more attractive. This work demonstrates how multi-source big geo-data can be combined to build comprehensive place-based representations for different groups of people. Highlights: A framework to characterize places regarding type, human activity, group and locale using multi-source big geo-data. Deriving place group information from individual behavior data of different groups of users from social media platforms. A data-driven approach is presented to uncover the inconspicuous-nice places of a city using street view images and check-in records.