Project information

  • Name : Recommendation System(content based)
  • Jupyter Notebook Viewer : Code
  • Aim : Build Recommendation System with Native approach. Trying to Implements RE without using ml algorithms.
  • Language : Python
  • Mathematics : One side relation compute
  • GitHub : Code

Motivation:

I love sci-fi movies, I watched 1000+ sci-fi movies and nowadays I am tired of IMDb recommendation.    
so I decided I want to build my recommendation system and it's different from the other recommendation.

`

Description

Let's Imagine What if we don't know any machine learning algorithm and we want to build a recommendation system then what is our native approach.
Below the image represent my native approach.

  • Sorry, your browser cannot display this SVG.
  • Concept: How to compress one movie's all dimension in one dimension and relate with other movies.
  • movie_map: Here each movie has own rank based on it's imdb and rotten tomatoes with respect to our movie.
  • Exmaple: Avatar : Pacific rim,Replicas (Based on rank).
  • Advantage: Here Each movie own rank so our movie doesn't affect on rank.
    Suppose if we using cosine similarity then rank affect by our movie.
    User does not feel same movie suggestion.
  • Disadvantage: Computationally Costly Table=N X N (N=movie)
  • GUI: Working Continue