BookTourist

Inspiration

We love books and we love to travel.

What it does

Shows you places from books you read, proposes cheap flights.

How we built it

We used:

• NamedEntityRecognizer for extracting places and characters from books
• Clustering for selecting relevant locations
• falcon for building REST API
• firebase as database
• kiwi API for flights
• ReactJS on the front, Python on the back
• GoogleMaps Places API for finding places from book in real world
• GoogleMaps Geocoding API

• Scalingo infrastructure

  Challenges we ran into

• extracting named entities in satisfactory time

• selecting relevant places

• using Google Maps API for the first time (on FE)

• using external APIs and combine data from them

• UX guy learnt basics of React

Accomplishments that we're proud of

• extracted places from book
• improved knowledge of NLP
• managed to implement risky idea

What we learned

• work under pressure of time
• we polished our programming skills
• design thinking
• new APIs

What's next for BookTourist

• better algorithms
• faster extraction of locations
• local database of books
• pipeline for processing new books
• access to database of payed books
• push notifications when book read by user is available in system
• show fragments of book
• user interactions

Built With

• css3
• falcon
• firebase
• github
• google-places
• html5
• javascript
• kiwi-api
• machine-learning
• openstreetmap
• python
• react
• rest-api