Web Dev Bootcamp TIL Day-33 (Django proj. Start & Recommendation Systems )

 

 

 

• Content-based Filtering: For a given movie, we can convert its genre and title to a scaled vector. If a user watches that movie, we will recommend to the user other movies with similar vector "directions".
• Collaborative Filtering: We can convert the ratings of users to a scaled vector ----> we can then recommend to a certain user user content based on similarity to other users
  • We can incorporate an additional layer w/ latent variables that take into account the relationship between user and keywords/genres
  • Latent Factor Model: 사용자와 아이템을, 잠재적인 차원(Factor)들을 사용해 나타낼 수 있다고 보는 모델이다.