• Login
• Admin

Tian, Yuan (2022) Inferring User Needs & Tasks from App Usage Interactions. PhD thesis, University of Nottingham.

Preview	PDF (Thesis with Minor Correction Completed) (Thesis - as examined) - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader Available under Licence Creative Commons Attribution. Download (2MB) | Preview
Preview	PDF (Cover Letter for Thesis Minor Correction) (Thesis - as examined) - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader Available under Licence Creative Commons Attribution. Download (103kB) | Preview

Abstract

Mobile devices have become an increasingly ubiquitous part of our everyday life, which are not only used for basic communication. Nowadays, the need for mobile services arises from a broad range of requirements including both single app usage (e.g., check on the weather) and complex task completion (such as planning vacation) which may lead to lengthy operations within distinct apps. Understanding how users interact with apps could provide us with great signals for profiling users and help service providers/app developers/smartphone manufacturers to improve user experience and retention. Therefore, in this thesis, we present work towards inferring user needs and tasks from their app usage interactions.

Firstly, we aim to better understand users' behaviour on using one particular app under different contexts. There have been many researchers proposed models for recommending the app user would use next proactively. However, less work has been conducted to enhance the app usage prediction when a new user comes whose information is insufficient for learning. Additionally, besides predicting which app users would use, we aim to further investigate if the app dwell time could also be modelled based on various user characteristics and contextual information. By conducting the comprehensive analysis and experiments, we demonstrate that users' next app and the time spent could be effectively predicted at the same time.

Other than effectively serving the individual apps that correspond to users' simple needs, we aim to further understand the high-level tasks within users' minds while engaging with different apps. We focus on identifying and characterizing tasks from app usage behaviour and then leveraging the extracted task information for improving mobile services. We first present an automatic method that accurately determines mobile tasks from users' app usage logs based on a set of features. Given the extracted tasks, we further investigate if there are common patterns that exist among all the complex mobile tasks. Finally, we demonstrate that the extracted task information could benefit user profiling in demographics prediction and next task prediction, especially when compared to the traditional app-based methods.

To summarize, in this thesis, we conduct a more comprehensive study on modelling users app usage behaviour. Additionally, we propose to set the stage for evaluating mobile apps usage, not on a per-app basis, but on the basis of users' tasks. Finally, we provide the initial steps in shaping future research on investigating whether and how the extracted tasks could be applied for improving mobile services.

Item Type:	Thesis (University of Nottingham only) (PhD)
Supervisors:	Zhou, Ke Milic-Frayling, Natasa
Keywords:	Mobile app, app usage behaviour, user modelling, user engagement, app usage prediction, demographics, mobile tasks, behaviour learning, machine learning, data mining.
Subjects:	Q Science > QA Mathematics > QA 75 Electronic computers. Computer science
Faculties/Schools:	UK Campuses > Faculty of Science > School of Computer Science
Item ID:	69478
Depositing User:	Tian, Yuan
Date Deposited:	19 Apr 2024 13:59
Last Modified:	19 Apr 2024 13:59
URI:	https://eprints.nottingham.ac.uk/id/eprint/69478

Actions (Archive Staff Only)

Edit View