Poster presentation of 'Socialoscope: Sensing User Loneliness and Its Interactions with Personality Types' for Graduate Research Innovation Exchange at Worcester Polytechnic Institute.
1. Abstract
Motive: Man is a social animal, who needs rewarding social
contact and relationships to make him feel comfortable.
When this need is not met, he feels isolated, leading to
thoughts of not fitting in, feeling empty and isolated.
Objective: This research investigates Socialoscope, a
smartphone app that passively detects the loneliness in
smartphone users based on the user’s day-to-day social
interactions and smartphone activity sensed by the
smartphone’s built-in sensors.
Background
• “The most terrible poverty is loneliness, and the
feeling of being unloved” - Mother Teresa
• Loneliness, is not the same as being alone. One can be
alone and very happy at the same time. Being alone, can
be experienced as a positive emotion.
• Effects of Loneliness: Increasing levels of stress,
lowering self-esteem, anxiety, panic attacks, drug or
alcohol addiction and depression, weakening of
immune system, increased sleep problems, increased
blood pressure, irregular heartbeat, increased chances
of stroke and cardiovascular disease.
• Hindrances in tackling loneliness: Social stigma, lack
of resources, lack of skilled therapists, misdiagnosis.
• Most susceptible population: Older adults and
international students. [3]
Key Contributions
• Correlation of smartphone features with questions
from the clinically validated UCLA loneliness scale [1].
• Extend the list of features explored by prior work on
smartphone loneliness and personality sensing by
including new internet and social media features.
• Explore whether smartphone sensed loneliness is
correlated with the Big-Five personality types [2].
• Synthesize machine learning classifiers that detect
lonely smartphone users, while factoring in differences
in personality types.
• Research, develop and evaluate the intelligent
smartphone app, which detects lonely users, while
factoring in differences in personality types.
Socialoscope: Sensing User Loneliness and Its Interactions
with Personality
Gauri Pulekar and Prof. Emmanuel Agu (Advisor)
Computer Science Dept., Worcester Polytechnic Institute
Features Sensed and Tracked
Big-Five Personality Types [2]
Conclusions
• Research and develop Socialoscope, an Android app that passively monitors
the social interactions of smartphone users in terms of calls, messages, social
media, Bluetooth and Wi-Fi devices, emails, browsing and thereby detects
loneliness levels factoring in the users’ personality type.
• Useful to old adults who face challenges in monitoring themselves and using
smartphones.
• Useful to busy users who do not have want to invest their time and energy in
monitoring their social wellness.
• Directly impacts smartphone social health monitoring.
Analysis
• Feature extraction: Smartphone features will be extracted from the sensed
data and analyzed for statistical correlations with loneliness and personality
types.
• Statistical analysis: The correlation coefficient and p-value of each feature
with questions from the UCLA loneliness questionnaire [1] and Big-Five
personality questionnaire [2] will be computed.
• Synthesize classifiers: The most correlated features will be used to build
machine learning classifiers that can detect the level of loneliness of
smartphone users.
• Developing the app: Machine learning classifiers will be used to develop an
intelligent Android app that can detect loneliness levels based on the
monitored data.
• User Acceptance: The app will be evaluated in a user study where users will
be surveyed to assess the app’s usability, acceptance and functionality.
Approach
• Passively monitor various user activities.
• Pilot study consisting of 20 subjects whose smartphone activity will be
automatically sensed and uploaded to Dropbox account for two weeks while
loneliness and personality questionnaires are administered simultaneously.
• One-time Big-Five personality questionnaire [2] will be filled out by subjects
which will help determine their personality type.
• Daily prompts to declare their level of loneliness will be given to subjects
using questions based on the UCLA Loneliness Levels [1].
References
1. D Russell (1996), “UCLA Loneliness Scale (Version 3): Reliability, Validity, and Factor Structure”, in Journal of
Personality Assessment, 66(1):20-40.
2. G Chittaranjan, J BlomDaniel, and Gatica-Perez (2011), “Who’s Who with Big-Five: Analyzing and Classifying
Personality Traits with Smartphones”, in Proc ISWC 2011, Washington, DC, USA.
3. A Ong, B Uchino and E Wethington, “Loneliness and the health of older people” in Gerontology.
4. “Funf Sensing Framework”, https://code.google.com/p/funf-open-sensing-framework/source/checkout