Multi-IDT Dataset

Citation Author(s):
Department of Computer Science and Engineering, University of South Florida
Department of Computer Science and Engineering, University of South Florida
Department of Computer Science and Engineering, University of South Florida
Department of Computer Science and Engineering, University of South Florida
Submitted by:
Tempestt Neal
Last updated:
Tue, 02/20/2024 - 10:45
Data Format:
0 ratings - Please login to submit your rating.


This dataset includes input dynamics (keystroke, touch, and mouse), affect data (physiological measurements), video, and text data collected from research participants aged 6 and older. The dataset includes data from a diverse set of participants, identifying as Asian, White, Middle Eastern or North African, Black or African American, and Hispanic, Latino, or of Spanish origin). Additionally, participants represent both iOS and Android users. Participants attended three separate sessions in the Cyber Identity and Behavior Research Lab (Department of Computer Science and Engineering) at the University of South Florida, each involving tasks performed on a Lenovo ThinkCentre M710 workstation and a OnePlus Nord N10 5G smartphone. In each session, various input dynamics are collected transparently. This includes fixed (participants type a provided phrase) and free-form (participants enter a phrase of their choice) keystroke dynamics, directed (participants navigate using the mouse based on provided tasks) and free-form (participants use the mouse as they please) mouse dynamics, as well as touch dynamics.

Tasks included:

Essay Composition: Compose a provided essay (approximately 500 words for adults and 150 words for children) on the desktop computer. Children engage with a shorter, child-friendly story instead of a lengthy essay for better participation and to minimize fatigue. Collected desktop modalities include fixed keystroke and free-form mouse dynamics.

Password Entry: Enter three passwords five times each on both the desktop and smartphone (GmxPV3L, Nv5PHS!8kP8, and jxK&5sDpwfE+U for adults, and schoolRocks, g@me&play, and GmxPV3L for children). These passwords are designed to vary in difficulty, incorporating special characters, numbers, and a mix of lower and uppercase letters. Desktop modalities include fixed keystroke and free-form mouse dynamics, while phone modalities include fixed keystroke and free-form touch dynamics.

Recipe Search: Search the web for a recipe of the participant's choice on the desktop. Collected desktop modalities encompass free-form keystroke and both directed and free-form mouse dynamics. Participants can opt to use a provided website for a guided search (directed mouse) or freely explore the internet (free-form mouse).

Mock Credentials: Generate mock username and password combinations as if setting up personal email, utility, banking, work-hosted email, and school-hosted email accounts on both the workstation and smartphone. Mock accounts for children include a streaming service (e.g., YouTube), a chat account for communication with friends, a banking or money-saving account, a gaming account, and a school account. Collected desktop modalities involve free-form keystroke and mouse dynamics, while phone modalities encompass free-form keystroke and touch dynamics.

Text Messaging: Engage in a brief conversation with a member of the research team via phone, simulating text messaging. Note that this task captures both input dynamics and the chat itself, and both components are incorporated into the dataset. Phone modalities collected encompass free-form keystroke and touch dynamics.The exchange is tailored to the participant's age group, addressing their study experience, covering topics such as:

  • Identifying the easiest and hardest tasks during the study.
  • Inquiring about their typical authentication methods on their own device.
  • Exploring reasons for not using any authentication method, including a password.
  • Understanding the participant's emotions when their chosen authentication method fails.
  • Gauging their opinion on continuous authentication approaches, such as those based on typing patterns.

To enhance participation across all three sessions, we offer flexible scheduling options for participants, allowing them to coordinate sessions based on their availability. Instead of scheduling sessions on set days, we send reminder emails after each session, prompting participants to schedule the next one. As a result, the time between sessions varies, averaging 13 days. The average duration of each session is 22, 19, and 19 minutes, respectively. This data collection has received Human Subjects approval from the University of South Florida's Institutional Review Board (STUDY002291). Participants receive compensation in the form of a $35 e-gift card per completed session.


To be added

Funding Agency: 
National Science Foundation
Grant Number: 



Submitted by Thibaud TRICHARD on Fri, 02/16/2024 - 13:05

Dataset Files

    Files have not been uploaded for this dataset