Introduction

This is the skeleton code of the backend API for mood sensing application. The backend service is implemented by Flask. Cassandra is the choice for database. I use google places API to find the nearest location given user's location. (https://developers.google.com/places/web-service/)

Assumptions

The mood data is very large and cannot be fit into a machine.
The third call API will return a nearest location (home, office, shopping center) if the given user is happy now. I assume the users are happy if the latest mood record in the database is happy and the timestamp is within one hour.

Define API

upload a mood: POST /api/users/<user_id>/ {"authentication": "XYZ", "location": {"lat": 33.3, "lon": 13.3}}

return the mood frequency: GET /api/users/<user_id>/moods {"authentication": "XYZ"}

return the proximity to a location: GET /api/users/<user_id>/proximity {"authentication": "XYZ"}

Layout code structure

main files are:

app/views.py: the backend API service.
unittests.py: test the backend service.
cassandraTable.cql: create tables in cassandra cluster. Define the partition key.

Design data model and key data structures

Three tables are stored in Cassandra. I choose NoSQL database because mood_fact table can potentially be very large. Cassandra can sort the records by time stamp, which is a advantage for proximity API.

mood_fact

user_id(PK)

date

time

mood_key(FK)
mood_dim

mood_key(PK)

mood
authentication_token

token

Unit tests