In this paper, we exploit different facets of the Friends and Family study to deal with two personality-related tasks of paramount importance for the user modeling and ubiquitous computing fields. First, we propose and validate an approach for automatic classification of personality traits based on the ego-networks’ structural characteristics. Our classification results show that (i) mobile phones-based behavioral data can be superior to survey ones for the purposes of personality classification from structural network properties and (ii) particular feature set/network type combinations promise to perform better with given personality traits. Then, we investigate the mediating role played by personality in the context of inducing behavioral change, specifically increasing daily physical activity using social strategies (social comparison and peer pressure). Our results confirm the role played by Extraversion and Neuroticism. Extroverts exposed to a social comparison strategy are positively associated with an increase in physical activity level, while they tend to decrease physical activity level if they are exposed to a peer pressure intervention strategy. Regarding Neuroticism dimension, neurotic people tend to increase their physical daily activity level if they are exposed to a social comparison strategy. Our findings may have implications in designing personality-based behavioral change strategies and suggest to incorporate users’ personality models in the implementation of persuasive systems.
12 March 2010 in Personal and Ubiquitous Computing
A diary approach was used to explore location-sharing preferences of 126 parents with young children with the goal of investigating which elements play a role in their decision to share their location. During a 3-week user study, we daily collected parents’ preferences of location sharing along with data related to the physical and social context, their interest in socializing with other parents and their emotional states. Our analysis points out several insights for this previously underexplored user group. In particular, our results suggest a relative greater importance of the context, both physical and social, with respect to individual traits such as personality, trust dispositions, and demographic characteristics. Moreover positive and negative emotions seem to influence the intention to share location information in a peculiar way: positive moods like happiness seem to encourage private sharing with selected people such as partner, relatives, friends and nearby parents, yet, when parents experience a negative mood, such as being worried, angry or sad, their intention to share publicly on the web is higher.