CRAWDAD unical/socialblueconn

Traces of Bluetooth encounters, Facebook friendships and interests of a set of users collected through SocialBlueConn application at University of Calabria

The dataset contains Bluetooth device proximity data collected by an ad-hoc Android application called SocialBlueConn. This application was used by 15 students at University of Calabria campus in Rende, Cosenza (Italy) and logged both the internal contacts between the participants and contacts with other 20 external mobile nodes. The dataset includes the social profiles (Facebook friends and self-declared interests) of the participants.  

date/time of measurement start: 2014-01-28

date/time of measurement end: 2014-02-05

collection environment: Experimental data were collected at the campus of University of Calabria in Rende (CS), Italy. 35 students were gathered in order to form the experiment team. After the presentation of the project, 15 students decided to participate to the experiment. During the experiment's lifetime, the 15 students detected also 20 external devices. Hence, the Bluetooth proximity data are relative to a total number of 35 devices. The experiment lasted one week during student's lessons, from January 28, 2014 to February 5, 2014, including only the working days.

network configuration: The network is a Bluetooth-based opportunistic network created among the participating devices. Each device performs a periodic Bluetooth device discovery every 180 seconds to find out about nearby devices. The smartphone's hardware of each device is different, with a Bluetooth range of about 10 meters; the operating system for all devices is Android.

data collection methodology: In order to gather the proximity information, an ad hoc application called SocialBlueConn was installed on each student's smartphone. Each participant was instructed to keep the device with himself and powered on from 12 a.m. to 20 p.m., and to use the application for mobile social networking during the experiment. Each device recorded the results of the periodic device discovery; all the wireless contacts were recorded in a text file on the device's SD. Each contact has a timestamp based on the device clock and reported as a relative time in milliseconds, referring to Thu Jan 01 1970 01:00:00 GMT 0100, the start of Unix Epoch Time. In order to obtain the Facebook friendships, the participants were asked to login with their Facebook credentials to an ad-hoc website accessing Facebook API. Once the students logged in, their friend lists and their social profiles were collected and sent to a central server. Participants' interests were collected at the beginning of the experiment through an offline questionnaire. The questionnaire contained a list of questions regarding participants' preferences according to the following macro-categories: (a) mobility, (b) sport, (c) music, (d) cinema, (e) literature, (f) multimedia entertainment, (g) politics, (h) other hobbies and (i) social networks. For each category, students could choose one or more sub-categories.

sanitization: Sensitive identifiers as Facebook identifiers and Bluetooth MAC addresses were replaced with random integer IDs.

limitation: The Bluetooth data communications suffer from the known limitations of the Bluetooth technology. The device discovery process was slow and regularly missed some nearby devices; links often failed when there were many Bluetooth devices in range. The timestamps among different devices are not synchronized. The clocks have been set manually to the same reference time at the beginning of the experiment.

Traceset

unical/socialblueconn

• file: contacts.zip, friendships.zip, interests.zip
• description: The dataset contains Bluetooth device proximity data collected by an ad-hoc Android application called SocialBlueConn. This application was used by 15 students at University of Calabria campus in Rende, Cosenza (Italy) and logged both the internal contacts between the participants and contacts with other 20 external mobile nodes. The dataset includes the social profiles (Facebook friends and self-declared interests) of the participants.
• measurement purpose: User Mobility Characterization, Social Network Analysis, Human Behavior Modeling, Opportunistic Connectivity
• methodology: Experimental data were collected at the campus of University of Calabria in Rende (CS), Italy. 35 students were gathered in order to form the experiment team. After the presentation of the project, 15 students decided to participate to the experiment. During the experiment's lifetime, the 15 students detected also 20 external devices. Hence, the Bluetooth proximity data are relative to a total number of 35 devices. The experiment lasted one week during student's lessons, from January 28, 2014 to February 5, 2014, including only the working days.

unical/socialblueconn Traces

• contacts: The Bluetooth contacts logs
• configuration: The trace records all the nearby Bluetooth devices reported by the periodic Bluetooth device discoveries.
• format: txt: ID1; ID2; Timestamp

ID1 represents the source device and ID2 the destination device. The timestamp is based on the device clock and reported as a relative time in milliseconds, referring to Thu Jan 01 1970 01:00:00 GMT 0100, the start of Unix Epoch Time. Bluetooth device discovery is asymmetric; a device A may see device B at some point in time but not the other way around.

• friendships: Facebook friendships between the participants
• configuration: Using Facebook API, we developed an application collecting the friend lists of each candidate.
• format: txt: In the file there is a square matrix, where the 15 nodes of the network are represented on the first row and first column. We define the existence of a friendship between nodes A and B through the presence of the value 1 on the corresponding position in the matrix. Due to the fact that the friendship relation is symmetric on Facebook, the matrix is symmetric.

• interests: Self-declared interests of the participants over different categories
• configuration: Participants' interests were collected at the beginning of the experiment through an offline questionnaire. The questionnaire contained a list of questions regarding participants' preferences according to the following macro-categories: (a) mobility, (b) sport, (c) music, (d) cinema, (e) literature, (f) multimedia entertainment, (g) politics, (h) other hobbies and (i) social networks. For each category, students could choose one or more sub-categories.
• format: txt: There are 10 .txt files.

  Interest_legend: it contains a legend in which the associations between the IDs and the major categories and subcategories are specified. Interest_MacrocategoryX (where X varies from A to I): it contains a matrix where on the first row there is the reference to the macrocategory, on the second row the references to the subcategories and on the first column the 15 nodes' IDs. We define the node's choice of a specific category and subcategory through the presence of the value 1 on the corresponding position in the matrix.