We were inspired by applications like Spotify, Tinder, and Ping (Museum app) which are helping the user to get recommendations based on their choices or ratings. We are implementing the application called YourGuide which will better the user experience of visitors to the museums or specific exhibitions walking by recommending art pieces that are similar to the users liking profile.
What it does
We are using a QRcode scanner to identify the objects in the first phase, afterwards the user gets a chance to give a rating for the specific object. During the rating process, whenever the user wants to have an art recommendation, our backend algorithm can recommend a list of art pieces that for the user to go next in the exhibition. This list is trimmed on the user's art profile and historic baseline data.
How we built it
We divided the project into three phases:
- Frontend UI
- Backend operations
Challenges we ran into
- Dataset - the dateset provided did not prove to align well with our use case
- AI model backend - our first approach, an unsupervised learning clustering did not prove to be useful in our use case, so instead we utilized a heuristic approach.
Accomplishments that we're proud of
Managed to get a team working that has not worked together before Manged to create an MVP
What's next for YourGuide
- User research for Frontend improvement
- Coordinate museum data to our system
- Collect other metrics e.g. time spend
- Have a more seamless user rating process, i.e. video OCR user input