Tips, Dones and Todos: Uncovering User Profiles in Foursquare
Total Page:16
File Type:pdf, Size:1020Kb
Tips, Dones and ToDos: Uncovering User Profiles in FourSquare Marisa Vasconcelos Saulo Ricci Jussara Almeida Universidade Federal de Universidade Federal de Universidade Federal de Minas Gerais Minas Gerais Minas Gerais Belo Horizonte, Brazil Belo Horizonte, Brazil Belo Horizonte, Brazil [email protected] [email protected] [email protected] Fabrício Benevenuto Virgílio Almeida Universidade Federal de Universidade Federal de Ouro Preto Minas Gerais Ouro Preto, Brazil Belo Horizonte, Brazil [email protected] [email protected] ABSTRACT 1. INTRODUCTION Online Location Based Social Networks (LBSNs), which com- Online Location-Based Social Networks (LBSNs) is a new bine social network features with geographic information paradigm of online social networks that has been experienc- sharing, are becoming increasingly popular. One such ap- ing increasing popularity, attracting millions of users. In plication is Foursquare, which doubled its user population LBSNs, users can check in, broadcasting their location to in less than six months. Among other features, Foursquare their friends, through the social graph. Check ins are per- allows users to leave tips (i.e., reviews or recommendations) formed in special locations, named venues, which represent at specific venues as well as to give feedback on previously physical places, such as universities, monuments, or even posted tips by adding them to their to-do lists or marking business enterprises and commercial brands. Check ins may them as done. In this paper, we analyze how Foursquare be converted into points that allow users to earn badges, users exploit these three features – tips, dones and to-dos venue mayorships as well as receive special offers. Exam- — uncovering different behavior profiles. Our study reveals ples of currently popular LBSNs are Foursquare, Gowalla 1 the existence of very active and influential users, some of and Brightkite , out of which Foursquare is perhaps the one which are famous businesses and brands, that seem engaged with the largest user population. Indeed, it has doubled in posting tips at a large variety of venues while also receiv- the number of users in only six months, having reportedly ing a great amount of user feedback on them. We also pro- reached the mark of 10 million registered users [25]. vide evidence of spamming, showing the existence of users In Foursquare, in particular, users may also post tips to that post tips whose contents are unrelated to the nature or specific venues, aiming at sharing information on any aspect domain of the venue where the tips were left. related to the venue with others. For instance, a tip left at a restaurant may suggest a special dish or even complain Categories and Subject Descriptors about the service. Tips can thus be seen as reviews or rec- ommendations, either positive or negative, about the tipped H.3.5 [Online Information Services]: Web-based ser- venues. A user who sees a tip may add it to her to-do list or vices; J.4 [Computer Applications]: Social and behav- mark it as done, as a sign of agreement as well as a feedback ioral sciences regarding the tip. As such, tips, dones and to-do lists are valuable features General Terms for fostering interactions among users, who can share their experiences and learn from others. Similarly, business own- Measurement, Human factors ers can also benefit from tips about their stores or prod- ucts, as these tips are a means for them to reach and get Keywords feedback from potential customers. Some businesses, which user behavior characterization, location based social net- are Foursquare users themselves, can also use tips for pro- works, spamming moting their brands and products [5]. Indeed, according to [26], around two thirds of the users post tips to Foursquare venues, showing the wide usage of the feature. Unlike user check ins, which propagate through the social links, and thus Permission to make digital or hard copies of all or part of this work for are visible only to the user’s friends, tips are visible to every- personal or classroom use is granted without fee provided that copies are body. Thus, tips have the potential to significantly impact not made or distributed for profit or commercial advantage and that copies online information sharing and business marketing. bear this notice and the full citation on the first page. To copy otherwise, to A few recent studies have analyzed the properties and user republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. 1 WSDM’12, February 8–12, 2012, Seattle, Washington, USA. http://www.foursquare.com, http://gowalla.com, and Copyright 2012 ACM 978-1-4503-0747-5/12/02 ...$10.00. http://brightkite.com/, respectively. characteristics of LBSNs, particularly Foursquare. However, mers in Foursquare is a hard task that is subject to a high most of these studies focus on the user check in dynamics, on degree of controversy. the properties of the social graph and related geographical The rest of this paper is organized as follows. Section information [17, 22, 18]. We are not aware of any previous 2 discusses related work, whereas Section 3 introduces the analysis of how users exploit the tip, done and to-do fea- main elements and features of Foursquare, our crawling method- tures. An investigation of relevant user behavior patterns, ology and a summary of the collected dataset. Section 4 an- when it comes to posting tips at venues and marking them as alyzes how users exploit tips, dones and to-dos, character- done or to-do, is key for future designs and developments. izing attributes of venues and of individual users, whereas Given that LBSNs, and Foursquare in particular, are be- the identified user profiles are presented and discussed in ing increasingly used as forums for recommending places, Section 5. Section 6 summarizes the paper and discusses services and products as well as for Internet marketing, un- possible directions for future work. derstanding how these features are used can produce useful insights into the potential effectiveness and vulnerabilities 2. RELATED WORK of these actions. To shed some light into the current use of tips, dones and Social interactions among individuals located within a short to-dos, we here present a characterization of user behavior physical proximity has been used to explain a number of in Foursquare. Our analyses are performed over a dataset phenomena in society, such as the proliferation of specific in- containing over 1.6 million venues and associated informa- dustries in a certain region [24] and individuals employment tion crawled from Foursquare during 8 weeks of operation status [29]. In the context of online social networks, Liben- (May to July 2011). For each collected venue, we gathered Nowell et al. [16] showed a strong correlation between friend- ship links and geographic location of those friends for Live- its tips, the users who posted each tip, the number of dones 2 and to-dos associated with each tip as well as the Foursquare Journal users. More recently, some articles demonstrated that geographically identified social content, like chatter from category to which the venue was assigned and its geograph- 3 ical location. Twitter , can be used to monitor real-world events and cre- Our study has two main phases. First, we characterized ate interesting applications. Particularly, Gomide et al. [10] venues and users with respect to number of tips, number proposed a spatio-temporal approach to identify potential of dones and to-dos as well as percentage of tips containing dengue epidemics, whereas Sakaki et al. [19] proposed to links (i.e., URLs or email addresses). Next, we applied a treat Twitter users as sensors and use them to create a mech- clustering algorithm to group users into profiles based on anism for earthquake detection of earthquakes. three attributes, namely, number of venues tipped by the However, there is still a limited set of studies about location- user, total number of dones and to-dos associated with the based social networks, possibly because they are still at an user’s tips, and the percentage of the user’s tips containing infant stage. Next, we briefly survey some of these studies. links. We further analyzed the clustering results by manu- Scellato et al [21] analyzed the social, geographic and ally inspecting the contents of the tips posted by samples of geo-social properties of four social networks that provide users randomly selected from each cluster. location information about their users, namely BrightKite, Our study revealed four different user profiles. Two pro- Foursquare, LiveJournal and Twitter. They showed that files correspond to users with different levels of activity in LBSNs are characterized by short-distance spatial-clustered the system: whereas one corresponds to occasional users friendships, while in the other types of networks, such as that post tips to only a few venues and receive only a few Twitter and LiveJournal, users have heterogeneous connec- dones and to-dos, the other, containing the vast majority of tion lengths. An analysis of Gowalla users showed that the the clustered users, consists of more active users who also number of friends follows a double Pareto-like distribution, tend to get much more feedback regarding their tips. A whereas the numbers of check ins and places are better de- third profile consists of users who are characterized by tip- scribed by log-normal distributions [20]. The authors also ping a large number of different venues and receiving a very analyzed the temporal variations of such distributions, ob- large number of dones and to-dos in return. Based on the serving that users tend to add new friends at a faster rate feedback received on their tips, these users, many of which than they give check ins and go to new places.