JD iCity Makes A Successful Appearance at AAAI 2019

In his keynote speech, Zheng explained that the concept of urban computing includes urban perception & data acquisition, big data management, data-driven analysis and services provided. Zheng said that the urban perception & data acquisition process is facing lots of challenges, including acquiring more data with limited resources; insufficient data due to the extreme lack of sensors in cities; and serious data loss issue due to faulty sensors or other factors. To address these challenges, Zheng proposed to unify different types of data in urban spaces, such as crowd flow data, traffic flow, air quality, weather data and social media data into unified spatial-temporal data types, and then optimize the management of those data utilizing multiple specially designed spatial-temporal indexes. Zheng gave an example of how these techniques help with detecting parking violations by mining the massive trajectory data of shared bicycles. When talking about the layer of data-driven analysis in cities, Zheng emphasized that the structure of spatio-temporal data is different from simple text and image data.