Ratings: Add a personal movie experience estimator

  • 1
  • Idea
  • Updated 6 years ago
  • (Edited)
  • 1
Hi,

I have an idea that could help people decide which movies to watch, which ones they would like/rate best.

If someone (User A) rates 50-100 movies, the system already has a good feeling about what kind of movies that person likes most. If it then compares the ratings User A gave to other users' ratings of the same movies, it can find people that gave similar ratings. This group has rated possibly a lot more movies than User A. Chances are good, that these ratings will be a good estimate how User A would rate those other movies. This could be a personal movie experience estimator, much more precise, than simply looking at gender and age.

Best regards,
Andras
Photo of Andras

Andras

  • 1 Post
  • 0 Reply Likes

Posted 6 years ago

  • 1
6 years ago
Photo of Dan Dassow

Dan Dassow, Champion

  • 16649 Posts
  • 18756 Reply Likes
Hi Andras,

What you are describing is a form of cluster analysis. This Wikipedia article provides a good tutorial on the subject.
https://en.wikipedia.org/wiki/Cluster_analysis

Unfortunately, what you propose would take a lot of computer processing time. There are literally millions of people who rate films on IMDb. In effect, IMDb would have to compare every pair of users to determine which groups exist and what group to which a user would be assigned. This process would have to be done on a regular basis, weekly or possibly monthly, since users continue to vote on films and other titles.

IMDb uses other proprietary techniques for their recommendation algorithm for "Recommended for you" which you can find on IMDb's home page.
  • 1

Powered by Get Satisfaction