Sentiment ChatBot

Inspiration

Suicide is a leading cause of death and depression is often unnoticed by many individuals. We hope AL, our chatbot, can help those who need someone to talk to regardless of how they feel.

What it does

AL delivers a unique and positive response to users after they share their responses.

How we built it

With the help of Stanford's CoreNLP sentiment classifier, we were able to detect the sentiment of a variety of texts, returning to the user: "Very Positive," "Positive," "Neutral," "Negative," and "Very Negative" along with a scale from 0-4 that's assigned to each sentiment. Depending on the sentiment of the text, the chatbot will return a geared message.

Challenges we ran into

A few challenges we ran into include figuring out how to implement CoreNLP, creating responses that feel natural, and doing parallel work.

Accomplishments that we're proud of

We're proud of producing a functional chatbot and watching our project bloom from an idea to a final product.

What we learned

We deep-dived into deep learning, we learned how to do parallel work, and as a bonus, we also learned ice cream is from China!

What's next for Sentiment ChatBot

Implementing a GUI so users can use it on a dedicated website, and creating our own sentiment analysis model.

Built With

• corenlp
• java
• machine-learning