Inspiration
The inspiration for CibusNU was the constant hunt for a good place to eat near campus. With a desire to find a fun and suitable solution to this, CibusNU was created.
What it does
CibusNU is used to find locations serving food nearby campus, with student review systems that lets you see other student's opinions on nearby food options, made by people in a similar situation to you, as students, rather than random people online with other review systems. CibusNU also has AI search features which can help you find specifics you may not be able to find with usual filtering methods.
How we built it
We initially started drafting with theoreticals - designed the interface in figma, worked on decoding protobuf OpenStreetMap messages, plotting amenities on a graph, to ensure that we had a suitable idea. When this checked out, we proceeded to build a Python FastAPI backend with SQLite and a C# Blazor Frontend.
Challenges we ran into
A major challenge was finding a unique identifier and the quality of our data. We overcame this by using Food Standards Agency data. Some data quality could be supplemented with the use of AI. While end-users may not understand some complicated OSM tags, an LLM (Google Gemini in our case) can understand these and provide end-user-readable descriptions, which we use in our project. Another issue we experienced was authentication, and we worked around implementing a complex authorization system by simply having users select how many stars they want to rate a shop, then are generated a unique review code that automatically opens in their University Outlook, thereby allowing them to send a review in almost equal time without requiring prior in-app authorization, since the vast majority of the target end user already has their University Outlook set up on their phone.
Accomplishments that we're proud of
We parsed megabytes of raw OSM data, decoding protobufs and having about 400 shops worth of relatively high-quality data in Egham, as well as surrounding Windsor and Staines. We made contributions to OpenStreetMap (mainly tagging Food Safety Agency IDs to shops) so that others may benefit from our work - this is another thing we are proud of - while OSM nodes may be unreliable (as OSM editors delete/merge, we found a way to uniquely identify shops nonetheless, through the Food Safety Agency Ratings ID. A near-native mobile app experience thanks to Progressive Web App technologies.
What we learned
Geodata requires significant considerations due to the variety of mapped structures - some shops marked as areas for example had no one coordinate so some adjustments had to be made to calculate distance.
What's next for CibusNU
In the future of the project, the AI features will be further developed to implement things that were currently cancelled due to time constraints. This includes a personalised recommendation by AI based on reviews you've made.
Log in or sign up for Devpost to join the conversation.