A chatbot is used to conduct an online chat conversation via text in lieu of providing direct contact with a live human. However, most of the chatbots we encounter feels lifeless, boring, and unattractive. In this project, we want to feel a genuine human interaction with our bots which supposed to be honest, informative, and entertaining.
What it does
ScuffedBot replies to your question and provides information about our beloved university, SFU, in its own unique ways.
How we built it
Using Python and its libraries, we created a machine-learning model, ScuffedBot. We use Neural Networks (specifically through TensorFlow's Keras) to train our model to give output depending on our user's questions.
Challenges we ran into
- Brain-storming and the delegation of work took up a lot of our time, and we weren't sure which idea to go with. Our initial ideas consisted of predicting the stock market crash, creating an internet safety website / bot.
- System configuration errors on different OS used (Windows / Mac).
Accomplishments that we're proud of
Using available tools to efficiently deliver to students in a short period of time. Coming up with a great user-friendly interface, that's easy to use!
What we learned
- Working collaboratively with others using GitHub.
- Natural Language Processing
- Deep learning techniques.
- Building models.
What's next for ScuffedBot
ScuffedBot can be improved using other machine learning techniques this includes training the model with more datasets.