[This is the first part of a three-part post that provides a high level overview of the digital music landscape where Mavrix and MySwar fits in.]
Recommender: specific type of information filtering system technique that attempts to recommend information items (movies, music, books, news, images, web pages, etc.) or social elements (e.g. people, events or groups) that are likely to be of interest to the user. – Wikipedia
Recommendation engines work as blend of many algorithms and approaches, to find similarities between what you find interesting , and what you may potentially find interesting. Often people use a Collaborative filtering model, or ‘wisdom of the crowd’ approach to generate lists of music, movies, news and other items you wouldn’t have come across in the mess of information around.
Recommendation services have evolved over the decades as I’ve tried to outline below
- The idea of collaborative filtering was derived, when developing an automatic filtering system for electronic mail called Tapestry, over at Xerox Palo Alto Research in 1992. They needed to handle the large amounts of email and messages posted to newsgroups. Users were encouraged to annotate documents , and these annotations could be used for further filtering.
- Grouplens began as a research group in the University of Minnesota where the students made a system to recommend Usenet News. It collected ratings from Usenet readers and used those ratings to predict how much other readers would like an article before they read it. This recommendation engine was one of the first automated collaborative filtering systems in which algorithms were used to automatically form predictions based on historical patterns of ratings. The research project would eventually spin out the Movielens project in 1997 and be featured in a Malcolm Gladwell column.
- Engineers from the MIT Media labs created a email-based collaborative music recommendation system called RINGO. The community around this project eventually became known as the Helpful Online Music Recommendation Service (HORM). In 1999, it eventually spun out into a company called Firefly which was acquired by Microsoft where it was killed suddenly.
- Amazon is well-known for its item to item recommendation system. All recommendations are based on individual behavior. Whether you like to buy something because it is related to something that you purchased before, or because it is popular with other users, you have a list of social recommendations – what other users bought, or personal recommendations-based on your purchase history.
- Netflix encourages subscribers to rate the movies they’ve viewed, and their CineMatch program recommends titles similar to those well liked — regardless of a film’s popularity at the box office.
- Google news serves a personalized news feed by assimilating the user’s genuine news interests as validated by click history and influences of local news trends, together with a collaborative filtering method. The result is that you view articles that align to your interests.
Music discovery is the new keyword on the digital block. To put it simply, an event of listening to a song by accident, having it play in your head, get you to like it and have you realize you want to hear it again is simplified to a website/app doing all that work for you. The music recommendation world today is vastly different from the Ringo email system where you rated some songs on an absolute scale and emailed it to the system, which would reply with songs/albums it thought you would like.
Let’s look at some awesome platforms that are driving this new experience in the second part of this post.
Pingback: Digital Music Landscape II: Discovery | Mavrix