-
Welcome message from chat bot and user asking for bus times. Information displayed from TransLoc API.
Inspiration
Google Knowledge graph is a powerful semantic web API that allows web scrapers to collect structured and machine readable data from unstructured data on the web, and it enables intelligent assistants like Google Now to answer questions about knowledge in a natural way. However, things like Google KG are rarely aware of specific knowledge about college campuses. They know when flights are coming, but not when your school bus is coming. Our chatbot aimed to assist students on discovering information on their campus via a simple messenger interface. It also allows students to share events to the campus community in a trusted environment.
What it does
FaceBook messenger bot that allows users to seek information about Duke University campus, such as restaurants, events, and bus schedules.
How we built it
- Python back-end with Flask framework
- Google App Engine with Cloud Datastore for hosting and database management
- Wit.AI for natural language processing
- BeautifulSoup web scraper to scrape Duke restaurant information off campus website
- TransLoc API to scrape current bus locations and calculate expected arrival time
Challenges we ran into
Training the bot to parse and accept a broad set of queries.
Accomplishments that we're proud of
Implementing on-campus default information that can be crowdsourced/expanded by users.
What we learned
How to create a Facebook messenger bot from scratch with Wit.ai. How to read/write to a cloud NoSQL database
What's next for Duke Information Chatbot.
Spreading the chatbot to users in order to crowdsource information and make the bot more useful over time.
Log in or sign up for Devpost to join the conversation.