An Early Personalized Recommendation System – Firefly

I’ve been taking a look at personalized recommendations systems recently, and one of the systems that I remember from earlier days on the web was for music, through a service known as Firefly. I’ve always wondered whatever happened to the service, and the people behind it.

I know that at one point, the company providing this service was partnered with Yahoo, and then it was later purchased by Microsoft. This personalized system was originally developed at MIT, and was incorporated into a business in 1995, by graduate students and Professor Patti Maes. Some of the technology developed by the company transformed into Microsoft’s Passport system.

Wired has a nice write up of the company in an article titled Firefly’s Dim Light Snuffed Out

In an article from this past May, Pitchfork also discusses some of those early days, when the professor turned to her students for some music recommendations because she didn’t like what was playing on Boston radio at the time, in Chris Dahlen’s Better Than We Know Ourselves. The article also looks at some more recent music recommendation systems.

A 1995 paper from those early days from Professor Maes and one of her students, Upendra Shardanand, looks at serving recommendations: Social Information Filtering: Algorithms for Automating “Word of Mouth” (pdf). Here’s the abstract from that paper:

This paper describes a technique for making personalized recommendations from any type of database to a user based on similarities between the interest profile of that user and those of other users. In particular, we discuss the implementation of a networked system called Ringo, which makes personalized recommendations for music albums and artists. Ringo’s database of users and artists grows dynamically as more people use the system and enter more information. Four different algorithms for making recommendations by using social information filtering were tested and compared. We present quantitative and qualitative results obtained from the use of Ringo by more than 2000 people.

A 2001 feature from Salon, Personalize me, baby, also takes a quick look back at Firefly, and discusses a new company delivering personalization to music, with Dr. Pattie Maes
as one of the members of its advisory board – Media Unbound.

Dr. Maes is still teaching at MIT, where she is the head of the Ambient Intelligence Group, who appear to be having a lot of fun doing research on a wide range of topics. A recent (2005) publication from the group is InterestMap: Harvesting Social Network Profiles for Recommendations (pdf). Here’s a snippet from the abstract of that document:

In particular, for this work, we harvested 100,000 social network profiles, in which people describe themselves using a rich vocabulary of their passions and interests. By automatically analyzing patterns of correlation between various interests and cultural identities (e.g. “Raver,” “Dog Lover,” “Intellectual”), we built InterestMap, a network-style view of the space of interconnecting interests and identities. Through evaluation and discussion, we suggest that recommendations made in this network space are not only accurate, but also highly visually intelligible – each lone interest contextualized by the larger cultural milieu of the network in which it rests.

Upendra Shardanand is supposedly busy working on a news service, with Jeff Jarvis, that is intended to be released some time this year, and will “release technology that identifies the most important stories and most ‘trusted’ versions — a computerized or computer-aided ‘editor.'” At least, according to a profile piece on Craig Newmark, where he discloses that he is acting as an advisor to the startup they are working upon. Somehow I expect a certain level of personalization to be involved.

I started this post because I wanted to write about a new service that I came across, Library Thing, which is still in beta, but may be interesting to someone who has more books than he knows what to do with. It seems like a nice way to inventory, tag, and get recommendations on books, as well as find others who are interested in the same literature. While reading about it, I had one of those “whatever happened to moments.”

I also came across an interesting looking new research project going on at the University of Maryland involving recommendations of movies. You can sign up to take part in the study, and become a member of this group at Filmtrust.

A paper associated with the research taking place is FilmTrust: Movie Recommendations using Trust in Web-based Social Networks.

With Filmtrust, I sort of feel like I traveled back in time to when I first came across Firefly.

Share

5 thoughts on “An Early Personalized Recommendation System – Firefly”

  1. Excellent article.
    Very happy to find some new articles and research.
    It’s been a while since i’ve read something that included Firefly …
    I’ll be very happy if you take a few minutes looking at my blog and at my recommendation company U.[lik]

    I will spend the night reading your stuff and still wondering how I’ve managed to miss this post.

  2. Thank you, leafar,

    I’ve subscribed to your RSS feed so that I can keep an eye out for new posts from you. I would like to spend more time reading through your blog – some very interesting stuff over there.

  3. thanks a lot.
    You have a specific RSS for english reader.
    My english post are mainly about recommandation and virtual Identity.
    I am currently posting an article with you as a starting point.

  4. Something I would like to change – I am not proficient enough in any languages other than English to write posts in anything else, so my site RSS feeds are in English only.

    I’m looking forward to your post.

    Thanks.

  5. Pingback: leafar

Comments are closed.