We were inspired by TravelPerk's proposal to build a platform that would allow travellers to connect with other travellers. We think that an activity recommendation system which matches people by their preferred activities would lead users to live unforgettable experiences while travelling. So that's what we did, or at least tried to.
It is at this point we would like to point out that we are for the first time experiencing the joys and struggles of building a full-stack app; with little to no experience with connecting databases, APIs and logic with front-end libraries, we plunged into the unknown and pushed onward through every challenge that we faced. This resulted in a very rewarding experience, as we saw that with enough effort and work we could learn a whole new workflow and framework from scratch.
Our app gets user's travelling information from a Taipy web app questionaire and uses a mongodb database loaded with user's travels to locate people who the user can meet in their destination based on their preferences. Then, with Ticketmaster's API, we request events in the city and with the help of a zero-shot classification AI (hugging face API), we categorize the event's appeal to then suggest meetings between users based on their preferences.
The app isn't fully implemented, as we run into trouble with Ticketmaster's API (the API seemed to ignore our filtering requests and we couln't filter by date), and as we had to learn Taipy from scratch, we couldn't figure out the quirks of dynamicaly rendering our app. Despite of this, mongodb's integration and the use of the hugging face API for categorizing events has been fully implemented and we have learnt a lot about using these resources.
To solve the problem with taipy and for the sake of the presentation, we have done a quick interface in streamlit.
Overall, we are very proud of a small, yet ambitious, project which required all of us to learn new skills and learn to cooperate to solve problems with our own ideas. Even though we couldn't ship anything, the prototyping fase of out project has been everything but boring.
Built With
- hugging-face
- mongodb
- pandas
- streamlit
- taipy
- ticketmaster
Log in or sign up for Devpost to join the conversation.