Clone wiki

Dataset: Coakley mobile / Home

Coakley Mobile Biometric Dataset

Access to this repository is provided under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. In short, the dataset is intended for research and cannot be used for commercial applications.

Publications making use of this dataset should cite:

  1. Michael J Coakley, John V Monaco, Charles C. Tappert. "Keystroke Biometric Studies with Short Numeric Input on Smartphones." Biometrics: Theory, Applications and Systems (BTAS), 2016.

Description

This dataset contains users performing several different tasks on Android mobile devices.

keyboard.csv contains text entry on a soft keyboard.

phonenumber.csv contains the phone number entry on a soft keypad.

gesture.csv contains input from users performing gestures to navigate a web page.

keys.csv contains the entity:keyname map for the keyboard dataset.

Population

In the first round of data collection, there were 21 users, with 10 graduate students and 11 undergraduate students. The 10 graduate students only provided keyboard samples, while the 11 undergraduate students provided both keyboard and gesture samples.

The phone number dataset contains entries from 92 users. There are 62 users in the Sept. 2014 dataset with about 40 entries each.

Soft keyboard dataset

The soft keyboard dataset contains 21 users entering a phone number and two differ text messages. Data was collected in 4 sessions over the course of two semesters, with approximately 2 weeks between sessions. Not every user completed every task and session lengths vary. A session is the time from the launch of the keyboard application (whenever the keyboard appears onscreen) to the time it is hidden. Tasks from extraneous input (such as changing the username or some other input besides the tasks defined) are labeled as "Unknown".

Phone number entry (11 key presses):

914 193 7761 Enter

Text entry 1:

Hello guys how are you doing? This is a test for the keyboard. Please respond ASAP.

Hi

I heard you attended the conference on semiconductors last week. I really look forward to seeing a report from you.

Regards John Doe

Text entry 2:

Hi babe, how are you doing? I’m testing this keyboard. Please respond ASAP.

Hi

Are you receiving my message? This is not the phone I usually use, so I want to be sure we keep in contact. Also, I need to see you soon.

Regards John Doe

Gesture dataset

The gesture dataset contains 11 users and was collected in 3 sessions over the course of a semester, with approximately 2 weeks between sessions. Only 1 user provided 1 session, and 10 users provided at least 3 sessions. Users were told to answer a series of questions which required the navigation of a web page. See /images for application screenshots.

Columns description

Note that not every column appears in both files.

Column Description
user user identity (MD5 of the username)
session session identity
keyboard keyboard layout
task the task being performed
device device model
timestart time the session started
action touch action (press or release)
dpi screen DPI (dots per inch)
entity soft key identifier
keyname soft key name
orientation screen orientation
pressure Android pressure (0 for none, 1 for normal pressure). Could be > 1
time time of the event in milliseconds
tool {major,minor} length of the {major,minor} axis the pointing device (finger)
touch {major,minor} length of the {major,minor} axis of the touch
pointer ID of the pointer (continuous gesture, or a sequence of motion events)
{x,y} {x,y} coordinates of the center of the touch event
{x,y,z} accel {x,y,z} acceleration sensor values
{x,y} dpi screen resolation in the {x,y} directions
{x,y} max screen size in the {x,y} directions
{x,y,z} rotation rotation sensor values
{x,y,z} gyro gyroscope sensor values

For more information, see the docs for the Android motion event

Feature extraction

How to run the feature extraction script: $ python features.py phonenumber.csv features

Updated