Skip to main content

Co-location Pattern Mining of Geosocial Data to Characterize Urban Functional Spaces...

by Arif Masrur, Gautam Thakur, Kevin A Sparks, Rachel L Palumbo, Donna J Peuquet
Publication Type
Conference Paper
Journal Name
IEEE International Conference on Big Data
Book Title
2019 IEEE International Conference on Big Data (Big Data)
Publication Date
Page Numbers
4099 to 4102
Conference Name
The 4th IEEE International Workshop on Big Spatial Data (BSD 2019)
Conference Location
Los Angeles, California, United States of America
Conference Sponsor
Conference Date

Spatial Co-location Pattern (SCP) mining continues to play a critical role in understanding the morphology of urban functional spaces of world cities. It requires a large amount of fine-granular data and computing efficiency to handle the combinatorial explosion of co-location patterns. To this end, this work has two main contributions - i) We showcase a novel approach to perform SCP mining to characterize intra-city scale structure of urban functionality or co-located activity patterns using geosocial Points-of-Interest (POI) vector data. ii) We present a generalized and optimized parallel/distributed SCP mining algorithm implemented on a Hadoop MapReduce system and demonstrate the utility of our approach using the city of Berlin (Germany) as an example. The SCPs tend to vary across Berlin's municipal boroughs and at different spatial scales. Our findings on Berlin's functional structure conform to existing urban geography models. Such a data-driven exploration of massive urban POIs using distributed computing is first of its kind and can help better understand the changing dynamics of urban functionality, as well as physical, and social network structure around the world.