Exploring Venue Popularity Using Location Based Social Network

INTRODUCTION:

Location-based services have attracted remarkable interests over the last a few years, thanks to the fast deployment of broadband mobile networks and the increasing prevalence of versatile mobile devices. Leading online social networks, such as Facebook and Google+ have embedded location based services as an important feature. Foursquare, one of the most popular location based social networks (LBSNs) enables users to explore nearby places (e.g. for specials and discounts) and network with their friends. Foursquare had more than 10 million registered users with 1 billion check-ins. However, it is unclear why certain venues become popular, e.g. why they attract a large number of visits (check-ins) or comments (tips) from users, and what characteristics popular venues usually possess. Understanding and answering these questions is crucial to many applications, including venue recommendation and targeted advertising in LBSNs.

VENUE POPULARITY ANALYSIS:

Foursquare users explicitly express their interests in venues in two ways, including checking in and leaving tips to venues. A venue being frequently checked in indicates that the venue is popular in a sense that lots of people visit it and like to announce their visits to their friends. A venue being frequently tipped indicates that people are interested in the venue and would like to share their experience with all other users. In the following, the number of distinct users (who checked in the venue), the number of check-ins, and the number tips are three key statistics to analyze how the venue popularity is affected.

CATEGORICAL ANALYSIS:

Figure 1. gives the numbers of venues in nine top level categories specified by Foursquare. Professional & Other Places (22.6%), Shop & Services (20.7%), and Food (20.1%) are the three largest categories. Let's analyze and compare the venue popularity across different categories, in terms of the total and per venue check-ins and tips. Let C be the total number of check-ins for all venues. The percentage of check-ins in each category, i.e., Pr(Ci) = C(Ci)/C, forms the check-in distribution among different categories. Similarly, the tip distribution representing the percentage of tips in different categories. Figure 2. shows these two venue popularity distributions and the distribution of the total number of venues. Observe that the third largest category, Food (C3), consisting of 20.1% of total venues, generates far more tips than other categories, i.e. 43% of all tips were left to Food venues. On the other hand, Professional & Other Places (C6) and Shop & Services (C8) are the two largest categories, with 22.6% and 20.7% of the total number of venues, but they only attract 13–16% of check-ins, and 7–14% of tips. Moreover, the Travel & Transport category (C9) consisting of only 9.5% of the total venues attracts the most check-ins (around 23.4%).

Figure 1: Venue Distribution of different categories

Figure 2: Venue and total venue popularity distribution over categories

PER VENUE POPULARITY ANALYSIS:

Consider the per venue popularity among different categories. For each category Ci, the average check-ins obtained by each venue, i.e., per venue check-ins, is counted as the total number of check-ins C(Ci) divided by the total number of venues n(Ci), i.e, r(Ci) = C(Ci)/n(Ci), which reflects the average ability of venues in Ci to attract check-ins. Similarly, the per venue tips for each category. In addition, for a venue v, define the user stickiness s(v) as the ratio between the total number of check-ins to v and the total number of distinct users who checked in v. We use the average user stickiness of venues from Ci to evaluate the average ability of venues in Ci to keep frequent/recurrent visitors. Figure 3. lists per venue popularity for each category in terms of per venue check-ins and tips, as well as user stickiness. Moreover, let R be the summation of r(Ci), then the normalized per venue check-ins is computed as r̄(Ci) = r(Ci)/R. Figure 4 shows the normalized per venue check-ins, tips, and the normalized user stickiness. From Figure 3 and Figure 4, Observe that the Travel & Transport (C9) category generates the most per venue check-ins, which is 1.7 times higher than the average of the second highest category, Art & Entertainment (C1). In particular, Los Angeles International Airport (LAX) has the most check-ins (i.e., 740,551 check-ins). Secondly, Food (C3) has the highest per venue tips, which indicates that Foursquare users are more likely to share their experiences of visiting food venues. Moreover, Residence (C7) and Professional & Other Places (C6) have the highest user stickiness, meaning that visitors to these venues tend to revisit the same venues often. On the other hand, the Food category has high per venue tips, but low user stickiness, so those venues attract users with fewer recurrent visits. This is easy to understand, since office, school, and church are three large subcategories of the category Professional & Other Places, and users checking in venues in these and Residence category tend to be people who live or routinely visit there, thus have higher stickiness. In comparison, venues in the Food and Travel & Transport categories tend to invite temporary visitors.

Figure 3: Ranking of per venue popularity

Figure 4: Normalized per venue popularity and user stickiness over categories

SUMMARY:

We investigate what drives the Foursquare venues to be popular, namely, attracting people to visit (check-in) or leave comments (tips) to them. The popularity of each venue is captured by the venue’s statistics, such as the number of check-ins, users, and tips. These results imply that in absolute value, the Travel & Transport (C9) and the Food (C3) categories attract the most check-ins, whereas the Food (C3) category dominates other categories in generating user's interests of sharing tips. The above analysis implies that Travel & Transport (C9) is the most popular category in attracting per venue check-ins, whereas the Food (C3) category generates the most per venue tips. The Residence (C7) and Professional & Other Places (C6) categories attract users with the most recurrent visits.

REFERENCES:

1.  A billion check-ins to foursquare venues. http://blog.foursquare.com/2011/09  /20/billion/
2. Foursquare. https://foursquare.com/.
3. M. Allamanis, S. Scellato, and C. Mascolo. Evolution of a location-based online social network: Analysis and models. In IMC’12, 2012. 
4.  Yanhua Li, Moritz Steiner, Limin Wang, Zhi-Li Zhang, and Jie Bao Exploring Venue Popularity in Foursquare. In Fifth International Workshop on Network Science for Communication Networks, 2013.