SoJo

SoJo, short for social journal, uses artificial intelligence and tracking to encourage healthy interactions and minimize loneliness

Comment

Inspiration

One of the feelings many people are experiencing during the coronavirus pandemic is loneliness. Even with our efforts to stay safe and save lives, our usual ways of seeing family, friends or just familiar faces have been put on pause. Many people may feel lonely from time to time but these short-term feeling are usually fleeting. According to MetalHealth.com, long-term loneliness is associated with an increased risk of certain mental health problems, including depression, anxiety, and increased stress. The impact of long-term loneliness on mental health can be very hard to manage.

What it does

Humans are social creatures, even with limited contact, being able to see a familiar is very important for our mental health. To help with this issue, we propose a new system to help us adapt with technology to keep those strong social networks. Our Project, “SoJo” which is short for social journal, aims to use artificial intelligence with connection tracking to keep people from feeling lonely.

The functionality of our app includes a scheduling algorithm to determine the optimal time to contact each other. In addition, we incorporated a Deep Learning Chat bot for those who may not want to directly talk to someone. The bot can listen and adapt to conversations while monitoring the overall mental health of users.

How We built it

It was deployed on IBM cloud services. We utilized the CleverBot API as our chatbot to communicate with our end users. The frontend was developed using React.

Challenges We ran into

We had original planned to utilize a tensorflow nmt model for our chatbot. The bot utilized a seq2seq architecture using a adam optimizer and a very large vocabulary size.

Tensorflow1.X and Tensorflow2.X have many cross compatibility issues with other packages such as Numpy. Getting the right environment to use was certainly a challenge.

Accomplishments that We're proud of

Deployment of the Application, implementation of a good number of features.

What I learned

I learned about JavaScript Promises and how to use JavaScript functional components. I also learned about environment management.

What's next for SoJo

The potential of having a smarter speech assistant is to have Mobile Deployment in addition to further training of an nmt model catered to our use case of communicating with our end users in an impactful way.

Built With

Share this project:

Updates