Docminer

Inspiration

Our team was inspired to build a tool to streamline scheduling and boost productivity for students. Course outlines contain data such as homework deadlines or exam dates in an unstructured format that limits their utility. With Docminer, we aim to solve this problem by harnessing the power of technology and making information accessible, organized, and actionable. By automating the process of extracting important dates and events from course outlines and other PDF documents, we hope to help students stay on top of their schedules and achieve their goals with peace of mind.

What it does

With Docminer, you can easily upload a PDF document and it will automatically create an ICS file for calendar events or summarize lecture notes. All members in the Discord channel can then add the events to their calendar apps with just one click. By having the bot integrated into Discord, students can help each other stay on top of their tasks and boost productivity, making organization and scheduling effortless.

How we built it

We chose to build Docminer as a Discord bot to allow students to add it to their existing servers. The bot uses Python along with libraries such as pdfminer and OpenAI’s API to leverage LLMs and select the desired information in a format that can be easily converted to ICS file calendar events.

Challenges we ran into

Our main challenge was prompt engineering to ensure that OpenAI’s model returned data in a predictable format for ICS calendar events. Determination of the important tokens for words such as “quiz” or “homework” enabled us to overcome this challenge by biasing these terms in the model’s response and ignoring other dates in the pdf document.