Project Details
There are three major components that ALJI blends together: Expressive Writing, Machine Learning, and Human-Computer Interaction.
Expressive Writing
Disclosure and reflection have had a fundamental role in many therapies to promote healing. Personal journals are an excellent way of promoting some degree of disclosure and reflection in the absence of a mental health professional. Clinical studies of expressive writing find that there are several health benefits that come at nearly zero cost [1] [2]. The additions that ALJI places on top of expressive writing may extend this further. For ALJI to act as a platform for expressive writing, it must first provide an acceptable space where authors can share deeply personal information. Transparency is the most direct way to build trust in software, so ALJI has been founded as an open-source project. This lets anyone view and scrutinize all portions of the software for any breaches of confidentiality. In addition, ALJI has been designed to never communicate over the internet, restricting ownership and access to solely the author of the personal journal.
Machine Learning
Machine Learning is the use of computers to learn a task by finding patterns and inferring information, instead of following explicit instructions. As a very simple example: a computer can detect that the word ‘tired’ was used ten times in one journal and two times in another journal of a similar length. From this pattern, it may infer (without certainty) that the first journal author is more tired than the second. A computer could understand very natural language when equipped with a much larger vocabulary and a more nuanced interpretation. If you are curious to learn more about Machine Learning, these two short, enjoyable videos are informative: here and here. A longer, more detailed video about the inner mechanics (for one specific type of machine learning) can be found here.
Human-Computer Interaction
Interaction design has a great effect on the effectiveness of a program. ALJI’s communication with the journal author must be centered around improving their well-being through non-judgmental feedback. This feedback is critical, since it is the primary addition that ALJI uses to go beyond acting as a basic personal journal. To align with a typical personal journal as closely as possible, feedback is kept minimal during regular use. However, when a mental health crisis is detected and confirmed, then a response is given to the author guiding them towards clinical mental health support systems. This response will be carefully crafted by mental health professionals to encourage the author to contact these support systems.
About Me
I am Patrick Sullivan, a graduate student within Virginia Tech’s Computer Science Department. I have a Bachelor’s in Computer Science along with a minor in Psychology. My research on the ALJI project is advised by Dr. Bert Huang (CS), Dr. Tanushree Mitra (CS), and Dr. Lee Cooper (Psych). There are four major passions that I have dedicated my life toward: teaching, computers, psychology, and music (in no particular order). I thoroughly enjoy mixing them together when I can.
[1] Lepore, S. J. & Smyth, J. M. (2002). The writing cure: How expressive writing promotes health and emotional well-being. Washington, D.C.: American Psychological Association.
[2] Pennebaker, J. W., & Chung, C. K. (2011). Expressive writing and its links to mental and physical health. In H. S. Friedman (Ed.), Oxford handbook of health psychology (pp.417- 437). New York, NY: Oxford University Press.