This project was inspired by the 1992 MS DOS game "Dr. Sbaitso".
What it does
Our project pays homage to the text-based chatbot game Dr. Sbaitso, while bringing his broken sassiness into the modern home.
How I built it
Through Google Home, users can connect to our server and speak with Dr. Sbaitso. Our AI generates responses to your voice commands through natural language processing enabled by the compromise and salient libraries, evolving the original text-based game to a conversational experience. The AI also utilizes api.ai to respond in a particular way to specific keywords.
Challenges I ran into
This project was particularly difficult, becuase it is difficult to create contextual responses with natural language processing, that work for all possible input. The original Dr. Sbaitso game purposefully avoids outputting resources that engage the users input because of these challenges. Additionally the natural language processing libraries aren't ideal, occasionally mistagging word like "chocolate" as the inappropriate part of speech, verb in this example.
Accomplishments that I'm proud of
We addressed a number of natural language processing quirks regarding english sentence structure. Our solution selectively implements two natural language processing libraries to play to the strengths of each.
What I learned
We learned about many of the difficult challenges in natural language processing and some of the solutions to these problems. None of our members had experience developing for Google Home, so this was a good opportunity to expand our skill sets.