This work aims to recognise people's interests from their daily behaviours. Recognising personal interests contributes to the production of personalised behavioural interventions, which would have more traction and motivational value when compared to general interventions. Typically, self-reporting methods are used to understand people's interests. These methods rely on people's perception; despite that, in most cases, interests are demonstrated in an individual's daily activity. Moreover, self-reporting tools are discrete and hence do not capture interest dynamics, which require continuous observation; attempts to overcome this weakness through longitudinal analysis can be highly intrusive and prone to memory and recall biases. Digital devices such as smartphones and wearables can overcome such limitations and hence have the potential to capture interests from daily behaviour in a continuous, longitudinal and unobtrusive (passive) manner. However, the daily routine is not only formed from actions motivated by personal interests. Instead, many of our daily actions are motivated by other reasons such as obligations and external rewards. Therefore, understanding the motives behind our daily activities is essential to distinguish behaviours driven by personal interests from those motivated by other factors. In this work, we create a framework for recognising personal interests using smartphones. We create an approach that first derives behavioural features of individuals' daily routines from their smartphones' data (digital phenotyping). Then, we employ knowledge of human motivation to (1) infer interests without recourse of asking (unobtrusive) and (2) adapt to newly developed interests. We have conducted real-world experiments to inform and assess our method. The conducted studies were designed to longitudinally and continuously observe behaviours while people undertake their daily life. Our results showed the advantage of basing the recognition of personal interests on motivational knowledge. Compared to baseline methods, our approach significantly improved the recognition of interests by an average of 62% with p
Date of Award | 1 Aug 2022 |
---|
Original language | English |
---|
Awarding Institution | - The University of Manchester
|
---|
Supervisor | Sarah Clinch (Supervisor) & Simon Harper (Supervisor) |
---|
Motivation-based Interest Recognition Using Digital Phenotyping
Ibrahim, A. (Author). 1 Aug 2022
Student thesis: Phd