EMAKI/README.md

Exploring Natural Language Processing Methods for Interactive Behaviour Modelling

Guanhua Zhang, Matteo Bortoletto, Zhiming Hu, Lei Shi, Mihai Bâce, and Andreas Bulling

Proc. IFIP TC13 Conference on Human-Computer Interaction (INTERACT), 2023, York, UK

https://link.springer.com/chapter/10.1007/978-3-031-42286-7_1

---

This repository contains the EMAKI dataset:

• user: User ID 1~39
• task: Text entry and editing (3), image editing (4) and questionnaire completion (5)
• trial: Questionnaire completion had four trials
• type: Action type, one of ['mousemove', 'mousedown', 'mouseup', 'keydown', 'keyup']
• timestamp: Timestamp of the current mouse or keyboard action
• X/Y: On-screen coordinates of the current mouse cursor
• value: Left (1) or right (3) if type is mousedown/up; keystroke content if type is keydown/up; nan/none otherwise
• resolutionX/Y: The resolution of user's screen when performing the online user study

As well as code for

• Byte pair encoding (BPE)
• The Transformer-based classifier

---

If you find our work helpful, please cite:

@InProceedings{zhang2023exploring,
author="Zhang, Guanhua
and Bortoletto, Matteo
and Hu, Zhiming
and Shi, Lei
and B{\^a}ce, Mihai
and Bulling, Andreas",
title="Exploring Natural Language Processing Methods for Interactive Behaviour Modelling",
booktitle="Human-Computer Interaction -- INTERACT 2023",
year="2023",
publisher="Springer Nature Switzerland",
address="Cham",
pages="3--26",
isbn="978-3-031-42286-7"
}