Modelling Application Switching during Everyday Mobile Device Interactions
Description: Computational modelling of human behaviour has practical applications in mobile human-computer interaction and beyond. E.g. being able to anticipate and predict when users are likely to switch from one application to another may help to develop tools that increase productivity while minimising potential distractions. However, this task is challenging, in particular in mobile settings, given the variability in user behaviours and mobile phone usage patterns.
In this project, we want to explore the feasibility of modelling application switching to explain human interactive behaviour. We will explore the theory of computational rationality [1] that hypotheses that humans act rationally with the objective of maximising a utility function. We will also investigate prior work that has applied hierarchical reinforcement learning [2] for task interleaving.
Goal:
* Investigate and understand the computational rationality theory
* Investigate existing cognitive models that can be used to model different functions of human behaviour (e.g. the EMMA model for modelling visual attention)
* Develop a model for application switching within the framework of computational rationality by applying methods from (deep) reinforcement learning
* A promising dataset for this task is the Rico dataset [3] for interaction mining
Supervisor: Mihai Bâce
Distribution: 40% Theory, 30% Implementation, 30% Evaluation
Requirements:
* Strong Python skills
* Background and prior experience in (deep) reinforcement learning
* Interest in cognitive models
Literature: Zhang, Xucong, Yusuke Sugano, Mario Fritz and Andreas Bulling. 2015. Appearance-based gaze estimation in the wild. Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).
[1] Oulasvirta, A., Jokinen, J., & Howes, A. 2022. Computational Rationality as a Theory of Interaction. CHI Conference on Human Factors in Computing Systems
[2] Gebhardt, C., Oulasvirta, A., & Hilliges, O. 2020. Hierarchical Reinforcement Learning Explains Task Interleaving Behavior. Computational Brain & Behavior.
[3] Deka, B., Huang, Z., Franzen, C., Hibschman, J., Afergan, D., Li, Y., Nichols, J., & Kumar, R. 2017. Rico: A Mobile App Dataset for Building Data-Driven Design Applications. Proceedings of the 30th Annual ACM Symposium on User Interface Software and Technology.