Hypermedia as the engine for route planning

Using hypermedia as the engine for route planning.

Presented at the API Strategy conference in Berlin

Transcript

1. Hypermedia as the engine for route planning @PieterColpaert
2. How far do you live from work?
3. Open Trip Planner with data by @openov
4. Or did you route for happiness? by @danielequercia
5. User agents need timely access to a mind- blowing amount of constantly changing datasets
6. Trade-off: load on server or agents? Data dumps Smart servers full query language of HTTP Dataset split in fragments Smart agents + server functionality helping certain use cases Read more at http://linkeddatafragments.org
7. A web-service for your app http://api.{myapp}/?from={A}&to={B}
8. You choose the functionalities http://api.{myapp}/?from={A}&to={B} &departuretime=2015-04-25T10:45Z &wheelchairaccessible=true &modes=plane,railway,bus,car ...
9. Data, services and apps Servers publishing Open Data web-services e.g., • a SaaS route planner: from → to • the closest station to your current location? • other micro-services end-user apps e.g., profile based routing + ticketing + ... $$$ $$$
10. Data, services and apps Servers publishing Open Data web-services e.g., • a SaaS route planner: from → to • the closest station to your current location? • other micro-services end-user apps $$$ $$$ If we enable this, adding something in between will be possible as well
11. User agents and servers Servers publishing Open Data Route planning agents $$$ Let’s talk about this
12. Publishing data vs. data services Data dumps Smart servers Data publishing (cheap) Data services (rather expensive) Algorithms as a service Dataset split in fragments Smart agents + server functionality helping certain use cases
13. What does road data look like? a path point 1 point 2 another path point 2 point 3 yet another path point X point Y
14. a connection What does transit data look like? departureTime + location arrivalTime + location a connection departureTime + location arrivalTime + location a connection departureTime + location arrivalTime + location a connection departureTime + location arrivalTime + location time
15. How does route planning work? time The Connection Scan Algorithm (CSA)
16. Resource ... Resource 3 Resource 2 Resource 1 Let’s fragment our dataset time nextPage nextPage
17. Resource ... Resource 3 Resource 2 Resource 1 Route planning agent needs X requests time nextPage nextPage
18. • Cacheable resources → Less server load → Higher availability • The agent performs the algorithm • Resources from different locations at once Advantages
19. Current state JSON-LD vocabulary at: github.com/opentransport
20. github.com/opentransport 1. Transform any GTFS dataset to connections 2. A server implementation creating and linking resources 3. A javascript library for route planning using connections To be launched by ~July
21. Any questions? @PieterColpaert