Inspiration
We were inspired by the recent advances in technology like ChatGPT and other language simulators. We noticed that there was a disconnect in the accessibility of such tools in terms of keeping records, so we made a website to manage this process.
What it does
Primarily, it keeps track of user-entered data by date and displays them (similar to a digital diary). For each text submission, NLP is used to analyze the most probable emotion. This result is then sent to ChatGPT for advice based on the emotion.
How we built it
Flask was used to develop the web app. Python functions were used to handle the backend logic. Co:here was used for NLP. We used the bootstrap framework for the front end.
Challenges we ran into
We faced challenges in training the NLP model with Co:here. We gathered around 20 samples each for 9 emotions for training. We also faced challenges in keeping track of user-entered records. This is shown on the website and organized by date, based on which user is logged in.
Accomplishments that we're proud of
We were able to utilize many different software applications and APIs to accomplish our goals in a working app. Some examples include: OpenAI, Co:here, Flask, Jinja, Bootstrap, Python, and Requests.
What we learned
We learned how to effectively develop a record platform that successfully applies NLP for a real world purpose, which is to hopefully help people with their mental health.
What's next for Emotional Chatbot
We want to expand the capabilities of the chat bot. Although it is able to infer the emotion of the user, we want to implement a recommendation system that gives the user a variety of options given that emotion. For example, restaurants, activities, or even memes.
Log in or sign up for Devpost to join the conversation.