Rotten Tomatoes: Recommmender Systems

The recommmender system that I chose to analyze was RottenTomatoes. It is a movie and tv show review aggregate site mainly, but is also used as a central where you are able to find local movies and showtimes, and movie news.

  1. Scenario Design

  1. The target users are people who would like opinions of the next movie to watch.
    According to similarweb.com, which is a website analysis site that studies website traffic demographics, Rotten Tomatoes’s main users are 65.37% male and 34.63% female. The largest age group that uses the site are 25-34 years old.

  2. Their key goals are for users to discover the next new movie to watch that they would be satisfied with. They would like to also to cater to advertisers who are in the majority movie studios that are promoting their new movies. As we can see from the above image, “Bones and All” is a movie that is likely a paid advertisement from the studio. Along with that, there are popular streaming movies and tv with their corresponding score.

  3. I can help them accomplish those goals by creating better algorithms, as the site has many way to gauge the likability of a movie, in order for customers to be satisfied with the movie that they choose to watch. I would help create algorithms that personalize movie tastes in order for future selections to be more accurate. Since there are 3 types of reviews that are given by the site, one could use that further as a model to cross reference with your tastes in order for a greater sense of satisfaction. The genre of movie that you prefer can also be included in this model.
    An example of this would be if a certain movie like “Avengers” which was universally praised and you didn’t enjoy. I would create a interface where you would be able to dislike or like that movie and cross reference that with critics who felt the same. Future recommendations then would be weighted heavily on critics whose tastes matched your own.

  1. A main feature of the website is something called the tomatometer which is their metric in gauging the quality of the movie. This is calculated by a simple proportion of good reviews to bad reviews, that of which is only seen after a minimum of 5 reviews of that movie have been posted. Additional metrics include a score from top critics (authors who have been specifically by rotten tomatoes for their reputation) and an audience score which grades the number of positive audience reviews to negative ones.

  2. Moving forward, although the sites display of tomatometer, topic critic score and audience score is great in giving me different measures of likability, It doesn’t truly represent my taste in movies. As explained above, there can be ways to improve the sites recommendation capabilites by personalizing the rating meter to your tastes more specifically. Another flaw that I can see from their model is the fact that the rating system only ranks the good movies to bad movies but is not specific enough to give an accurate rating. Under this system, a majority or even all of the critics would be able to rate the movie a mediocre 3 out 5 and the movie would still receive the coveted “Certified Fresh” rating even if most people thought that the movie was mediocre. I would further try to weight their actual review into the tomatometer to give a more accurate rating.

Sources:

rottentomates.com

similarweb.com