Starting with the Recommender
We made it to Day 1 of 80. In Day 0, I announced my project, TVLens. Currently, we have the landing page and newsletter set up for the project.
Now for Day 1, the recommender is the heart of TVLens, so I want to build it first. I want to spend more time working on it than the UI/UX. Focusing on the recommender first also helps me learn a bit about algorithms.
The Recommender
There is nothing complicated about the recommender right now. The v1 of the recommender links shows that share a cast member. I wrote a query that takes a show, finds all its cast members, and then finds every other show each cast member appears in. I rank the results by how many cast members each show shares with the one I started from.
In today’s test it worked. Breaking Bad returns Better Call Saul and Grey’s Anatomy returns Private Practice. These are TV shows someone would mention if they were talking about an actor or actress in both TV shows.
Observations
The data was enough to prove the cast query works, but only 37 of my 100 shows came back with anything. I suspect this is because the cast is truncated. I will look at the TMDb ingested data in the future.
Next Up
Actors are one type of relationship. The next type of relationship I want to explore in this graph of TV shows is crew members. Crew members show up across more shows than cast members do. For example, I found that Dick Wolf shows up across the Law & Order universe and in the Chicago franchise.
See you tomorrow for Day 2!