Inspiration
We realized that we had an opportunity to create a great experience for rising college freshmen regardless of visual acuity. College is supposed to be a place for everyone, so we figured the resources for those interested should be more inclusive as well.
What it does
College Chat program is a speech based chatbot that provides prospective University of Maryland students with any information they may be looking for regarding programs and admission.
How we built it
We built the program around the use of AzureOpenAI with Python and Javascript. We were able to create a speech interface that remembered queries and implemented RAG in order to bolster the ChatGPT 3.5 LLM for our purposes. In order to implement RAG, we scraped a variety of UMD links and APIs (specifically umd.io) to populate a database for document retrieval.
Challenges we ran into
Some challenges we ran into were getting speech to transfer to a web application as well as setting up our OpenAI resources to begin the project.
Accomplishments that we're proud of
We're proud of developing a chatbot that people can use regardless of any potential visual impairment as well as for utilizing RAG and effectively training our model with diligently sourced data concerning the University of Maryland.
What we learned
We learned a great deal over the course of the project, including how to make use of LLMs, scrape data using APIs, and implement speech interfaces.
What's next for College Chat
The next step for College Chat is forming a new name. Also, we want to expand to a UI. We had some difficulty integrating audio to the website. We
Built With
- azure
- database
- javascript
- openai
- python
Log in or sign up for Devpost to join the conversation.