Kwery

How It Started :D

As students with plenty of experience navigating through lionpath, schedule builder, the undergrad bulletin and a plethora of college and department specific websites, we've grown familiar with the frustrations that come with planning your academic journey. University resources are spread out and redundant, often making a simple question a 30 min egg hunt on the web. Hence, we set out to develop a way to streamline the advising process and equip students with an easy, user friendly and enjoyable tool to plan their academic journey. Or find out when the gym is open! Kwery knows it all.

What it does... for now!

When prompted, Kwery generates a response based on dataset of publicly available information for most of Penn State's University resources that have a website. That includes the undergrad bulletin with all majors (and their subsequent options), minors, classes and their respective pre-requisites. Additional resources currently include CAPS. Kwery knows undergraduate major graduation requirements, all undergrad class pre-requisites and content, prospective career information and can advise on such.

How we did it

A significant portion of Kwery's operations happen in Python files. It parses through a PDF of the university bulletin, feeding information to ChatGPT 3.5 in increments of 3,000 words. The output is a large table of questions and answers. When a question is fed to the next program, it uses a vector-search function to find the most relevant question, and outputs the corresponding answer.

Challenges we ran into

The vector-search function in Python has a limit to the number of entries it can search through, and this limit happens to be significantly lower than the number of questions in our table.