Echo Nest are claiming they can predict your taste in films based on your taste in music. They studied user accounts across various music streaming services and looked for correlations to film preferences. Using more than simple collaborative filtering, they mine sources which not only track what you listen to, but also when you search, skip or share music, plus what people are saying about that music. This analysis creates your “Musical Identity”, your personal taste and behavioural profile. They found some, perhaps obvious, correlations between rom com fans and a preference for artists who sing about relationships, such as Céline Dion. They also found that Jay-Z fans are into Get Rich or Die Tryin’ and Scary Movie 4.
This is a great combination of domain knowledge, as there is obvious crossover between film and music domains; musicals, film soundtracks, films about music and music about films, plus actors turned artist and vice versa. We’re hoping Echo Nest develop this work further and we see a real life implementation soon. It is an exciting step forward in Big Data, one we are keen to explore ourselves.
Such cross domain sources for recommendations are ripe for investigation. Other start ups such as Tastebuds have tried to match up potential dates and friends based on Spotify user profiles. Yahoo are also working on augmenting TV and film recommendations using browsing history of web pages or news stories. It's all getting a bit Big Brother, and we like it.