Ever wonder how the Netflix suggestion engine works? Each film is tagged with at least 100 data points. These tags are created not by metadata elves, but by people like Jordon Canning described in this article.
She watches movies and enters data into a spreadsheet on a laptop tracking everything from the characters occupations to the amount of smoking in the film. Some are straightforward and others are subjective -- how violent was the movie on a scale of one to five?
While this certainly clears up how the data is created, it doesn't clear up some of my lingering questions such as why certain films show up in my recommendations (algorithms are wacky) or why IMDB and Netflix don't join forces or integrate data.
No comments:
Post a Comment