Abstract
Human mobility mining has attracted a lot of attention in the research community due to its multiple implications in the provisioning of innovative services for large metropolises. In this scope, Online Social Networks (OSN) have arisen as a promising source of location data to come up with new mobility models. However, the human nature of this data makes it rather noisy and inaccurate. In order to deal with such limitations, the present work introduces a framework for human mobility mining based on fuzzy logic. Firstly, a fuzzy clustering algorithm extracts the most active OSN areas at different time periods. Next, such clusters are the building blocks to compose mobility patterns. Furthermore, a location prediction service based on a fuzzy rule classifier has been developed on top of the framework. Finally, both the framework and the predictor has been tested with a Twitter and Flickr dataset in two large cities.
Highlights
One of the most important social phenomena of the last decades has been the endless transference of population from rural areas to urban ones
In order to test the feasibility of our approach, we have developed a prediction service able to estimate where an Online Social Networks (OSN) user is going to post his or her document making use of the learned mobility patterns
The study of human dynamics is paramount for the development of innovative services in the context of large cities
Summary
One of the most important social phenomena of the last decades has been the endless transference of population from rural areas to urban ones. Current mechanisms focus on extracting general mobility information related to a particular urban area without distinguishing the time of the day in which the information was generated They do not study the relationship between the moment of the day at which social-media documents are posted and its associated spatial place. The usage of the spatial and textual content of OSN data makes the resulting model provide the location of the social areas of the city and a set of labels associated with each cluster describing its predominant activity or landmark giving rise to the most valuable information. In order to study the feasibility of the proposal, a lightweight location predictor has been developed on top of the proposed framework This service profits from people displacement between clusters in different time slots so as to forecast the location where an OSN user is going to submit his document.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
