#Recommendation Service API

Description

Short Akka implementation of a recommendation engine, full typelevel stack version coming soon

Installation instructions and running:

1. Download and Install the following:
   • Scala 2.13.13
   • SBT Version 1.3.13
   • Postgresql 15.4
2. run chmod +xw ./scripts/install_db.sh
3. run ./scripts/install_db.sh
4. to run tests use sbt -Dconfig.file=src/test/resources/application-test.conf test;
5. If you want to use the DB please insert data into movies and tags tables
6. run sbt run in your terminal at the root of this project
7. use the api contract in docs/openapi.yml to aid you in calling this api

Design decisions:

• Used Akka http to serve a restful API
• Cats allows us to easily compose functions together (with monads) and take advantage of category theory.
• Dependency injection allows for easier component construction and injecting dependencies for classes when required.
• Refined types could be used for improved type level validation but I did not have time for this
• Uses Postgres DB for optional storage of movies
• Score by tags runs into a problem where if you match a tag exactly e.g "Crime" = "Crime" it is given a score of 2 and when it doesn't match exactly e.g "Crime" = "Crime Family" it gives a score of 1. So when you get e.g "Crime Family" != "Crime" you get 0. I think this is fine because as a business has put "Crime" and "Crime Family" in the list of tags for one film. Meaning "Crime Family" comes under the umbrella of "Crime" but not the other way around, I think this is reasonable reasoning to why this algorithm remains accurate within business means.
• Also implemented a check on similar movie lengths
• From the given data points are don't think there is more refinement that can be done. Matching on title would not be a good Idea for a recommendation engine. That is like saying Aquaman is similar to Alvin and the chipmunks because they both have the Char 'A'