Friday, May 09, 2008
Startup Uses Meta Data for Movie Mining
The company received $1 million in angel funding, according to Mashable. Netflix and Blockbuster have been grappling with this challenge for some time but have yet to overcome it to date.
Movie and musical tastes vary greatly, and individuals can be fickle in liking some actors white not liking others.
Also, many movies sound great by their premise, the trailers, or the actors or directors, but then are panned by critics and viewers.
Solving this problem requires drilling down on the right data. Categories of mystery, sci-fi or romantic comedy are too general. I'm betting that other factors are far more indicative of an individuals preference. The intensity of the violence and profanity, pacing, appearance of nudity will say more about a movie than the genre.
The shift is focusing on the right data, and it should also include more direct questions rather than making assumptions based on characteristics that may not be relevant. Netflix should ask why we rated a movie 1 star. Was it the plot, actor, script, or humor? This would result in a better recommendation engine than guesswork.
By John Gartner at 11:03 AM | Comments (0)