MultiContext Free-Web Gaze Dataset (MCFW-Gaze) Dataset Description The MultiContext Free-Web Gaze Dataset (MCFW-Gaze) is a high-resolution eye-tracking dataset designed to capture human visual behaviour across multiple interaction contexts, including controlled laboratory tasks and real-world web browsing scenarios. The dataset contains continuous raw gaze recordings collected using a Tobii Pro Fusion eye tracker (120 Hz). It combines structured experimental tasks with unconstrained free-viewing conditions, enabling research on visual attention, authentication behaviour, human–computer interaction, and gaze-based modeling. Key Features Multi-context design:Includes natural image viewing, gaze-based authentication, password entry tasks, and free web browsing. Real-world web interaction:Participants engaged in extended (600 s) browsing sessions on real websites (Amazon, Yahoo! JAPAN News, YouTube), reflecting ecologically valid visual behaviour. High-resolution recordings:120 Hz binocular gaze data including 2D screen coordinates, 3D gaze origin, pupil diameter, eye openness, and validity flags. Continuous raw data format:All recordings are provided as time-series TSV files exported from Titta (PsychoPy wrapper for the Tobii SDK). Multi-session natural image task:100 images presented across three sessions (80 images included in this release due to copyright constraints). Experimental Contexts The dataset consists of recordings from 15 participants (ages 21–42). Stimuli were presented on a 1920×1080 display (31.0×17.5 cm) at an approximately 65 cm viewing distance. Four main task categories are included: Natural Image Viewing 5 seconds per image 100 images (80 released) Three sessions per participant Gaze Pattern Authentication 3×3 grid-based pattern reproduction Five predefined patterns User-controlled duration Password Entry (On-Screen Tenkey) Five blocks 6-digit numeric passwords User-controlled duration Free Web Browsing - Shopping: Amazon product search task (600 s) Free Web Browsing - News: Yahoo! JAPAN News browsing (600 s) Free Web Browsing - Video: YouTube video viewing (600 s) Data Contents Each sample includes: Device and system timestamps Normalized gaze coordinates (display area) 3D gaze origin and gaze point in user coordinates Head position within tracking box Pupil diameter Eye openness diameter Validity and availability flags All data are provided in tab-separated value (TSV) format. dataset/├── data/│ ├── participant_001/│ │ ├── image_01_1.tsv│ │ ├── image_01_2.tsv│ │ ├── image_01_3.tsv│ │ ├── gaze_pattern_auth_trial1.tsv│ │ ├── gaze_pattern_auth_trial2.tsv│ │ ├── gaze_pattern_auth_trial3.tsv│ │ ├── gaze_pattern_auth_trial4.tsv│ │ ├── gaze_pattern_auth_trial5.tsv│ │ ├── password_experiment.tsv│ │ ├── password_experiment_block1.tsv│ │ ├── password_experiment_block2.tsv│ │ ├── password_experiment_block3.tsv│ │ ├── password_experiment_block4.tsv│ │ ├── password_experiment_block5.tsv│ │ ├── shopping.tsv│ │ ├── news.tsv│ │ └── video.tsv│ ├── participant_002/│ │ └── ...│ └── participant_015/│ └── ...├── stimuli/│ ├── 20.jpg│ ├── ...│ ├── 99.jpg│ ├── password_example.png│ └── pattern_example.png├── participant.csv└── README.md Potential Applications MCFW-Gaze supports research in: Visual attention modeling Gaze-based authentication and security Human–computer interaction (HCI) User behavior analysis in web environments Biometric modeling Machine learning on gaze time-series data Ecological eye-tracking research Contribution MCFW-Gaze bridges the gap between tightly controlled laboratory eye-tracking experiments and ecologically valid real-world web interaction. By combining authentication tasks, structured stimuli, and free browsing sessions within a single unified dataset, it enables cross-context analysis of human gaze behaviour.