What do your food & drink habits tell about your culture?

Traditional ways to study cross-cultural differences depend on surveys, which are costly and do not scale up. We reveal another way to obtain similar data that could revolutionize the study of global culture.

We propose the use of publicly available data from location-based social networks (LBSNs) to map individual preferences. This is interesting because an LBSN check-in expresses the preference of a user for a certain type of place. LBSNs have also the characteristic to be accessible (almost) everywhere by anyone, solving the scalability problem and allowing data from the entire world to be collected, at a much lower cost (compared to traditional surveys). 

Users expressing preferences
Users expressing their preferences in LBSNs.

Our goal is to propose a new methodology for identifying cultural boundaries and similarities across populations using data collected from LBSNs. Since we know that food and drink habits are able to describe strong differences among people, we use Foursquare check-ins in such locations to represent user preferences for specific types of food and drink. We studied how these preferences change according to time of day and geographical locations. We have found that:

  • The eating and drinking choices in different countries, cities, or neighborhoods of a city reveal fascinating insights into differing habits of human beings. For instance, preferences among people in cities located in the same country tend to be very similar;
  • The time instants when check-ins are performed in food and drink places also provide valuable insights into the cultural aspects of a particular region. For example, whereas Americanss and English people tend to have their main meal at dinner time, Brazilians have it at lunch time.

Given those observations, we consider spatio-temporal dimensions of food and drink check-ins as users’ cultural preferences. We then apply a simple clustering technique to show the “cultural distance” between countries, cities or even regions within a city. We found that:

  • Our results often strongly agree with common knowledge;

  • Comparing our results with the World Values Surveys (a very large study based many years of survey data), the similarities are striking.

Clusters by cities
Clustering cities.

Yet, unlike traditional survey-based empirical studies, such as the aforementioned one, our methodology allows the identification of cultural dynamics much faster, capturing current cultural expressions at nearly real time, and at a much lower cost.

For more, see our full paper, You are What you Eat (and Drink): Identifying Cultural Boundaries by Analyzing Food & Drink Habits in Foursquare.

Thiago H Silva, Universidade Federal de Minas Gerais, Brazil
Pedro O S Vaz de Melo, Universidade Federal de Minas Gerais, Brazil
Jussara M Almeida, Universidade Federal de Minas Gerais, Brazil
Mirco Musolesi, University of Birmingham, UK
Antonio A F Loureiro, Universidade Federal de Minas Gerais, Brazil

About the author

Thiago H Silva

Thiago is a Ph.D. student in Computer Science at Universidade Federal de Minas Gerais (UFMG), Brazil, and passionate about cities, cultures, and people. During his Ph.D. he had the opportunity to perform cool research in different locations around the world.

View all posts