Social networks acquire data on most aspects of the daily life of millions of people around the world. Analyzing this rich and ready source of information can help us better understand the complex dynamics of society. In a recent item Posted in EPJ Data Science the authors propose the use of location-based social networks to study the activity patterns of different gender groups.
Guest posts by Willi Müller and Thiago H. Silva
Gender differences are subjective in nature and can vary greatly from one culture to another, making them difficult to explain. Indeed, in the past few decades, this topic has received a lot of attention from researchers, but there is still a long way to go to reach a consensus on the subject.
Traditional methods of studying differences between gender groups depend on surveys, which are expensive and do not increase. Furthermore, data produced under such conditions is commonly released after long intervals of time (for example, several years). Therefore, studies cannot quickly capture changes in the dynamics of societies. Furthermore, the results of studies of interregional gender differences are usually only available for large geographic regions, often countries. Thus, even though survey-based studies might be conducted in arbitrarily small regions, such as a city, a neighborhood, or even a particular place (for example, a university or a shopping mall), information on gender differences at a fine granularity space so fine are not readily available.
We present in another way obtain and explore similar data that could help the study of global gender differences. To map individual preferences, we propose to use publicly available data from location-based social networks (LBSN). This is interesting because when specific LBSN users check-in at a specific location they express their preference for this type of location. LBSNs are accessible almost anywhere by anyone and therefore allow the collection of data from all over the world at a much lower cost than traditional surveys.
We propose a new methodology to quantify differences between male and female preferences for venues in different regions at different spatial granularities, worldwide, based on LBSNs. The aggregation of these differences across multiple locations could then be used, for example, in constructing an indicator of gender differences in a given region. We illustrate the use of our methodology by extracting user preferences for venues located in different urban regions around the world from check-in data collected by Foursquare. We found that:
- Gender and place preferences may not be independent in specific regions. The level of geographic detail we analyzed varies from country to city, neighborhood and even individual location;
- Our approach could capture some essential aspects of gender differences. By comparing our results with an official gender differences index, we found evidence motivating the study of new approaches using LBSN data alongside other datasets in future developments of gender differences indices.
Our methodology could be a promising tool to support large-scale gender preferences for place studies that require less human effort and time, compared to traditional methods, and can react quickly to real-world changes because it is based on LBSN data . With this method it is possible to analyze gender cultural preferences for places, opening opportunities for different studies and applications in different areas.
Willi Müller is currently a graduate student at the Hasso-Plattner-Institute in Potsdam, Germany. In 2014 he visited the Federal University of Minas Gerais for one year, where he began to collaborate with the other authors of this work. Before that, he received a B.Sc. in IT-Systems-Engineering in 2013 at the Hasso-Plattner-Institute.
Thiago H. Silva is a professor at the Federal University of Technology, Paraná, Brazil. He holds an M.Sc. (2009) and a PhD. (2014) in Computer Science from the Federal University of Minas Gerais. Thiago has strong experience in industry and academia in the areas of urban and social computing. Learn more about: dainf.ct.utfpr.edu.br/~thiagohs