Reflection on GroupLens

The paper “GroupLens: An Open Architecture for Collaborative Filtering of Netnews” was presented by Paul Resnick, Neophytos Iacovou , Mitesh Suchak, Peter Bergstrom, John Riedl.

Reflection:

The paper presents the idea of GroupLens, Which at first glance reminded me of current “netnewsReddit. We are currently working on Reddit data for our project, so the paper really hits home. The use of the word lurker made me chuckle while reading the paper. The feature of divided the news into newsgroup is identical to subreddit and newsgroup being moderated by… well moderators, is still present in practice, and used in Reddit, but rather than it being done by a human all the time, the usage of “bots” have increased greatly. The biggest selling point of GroupLens was the prediction system Better Bit Bureaus, that’s based on the idea, “people who agreed in their subjective evaluation of past articles are likely to agree again in the future“. This is the core idea behind machine learning algorithms of recommendation systems that are used in modern times. Another feature voting is also similar to the Karma system of Reddit.

To be honest, I’m pretty impressed to see that, even in 1994, people came up with such a novel idea for recommendation system. They used a very simple version of it, but it seems like the building block of modern recommendation system, using user review, predicting the score of each news and recommending users based on the scores. I know I am bringing up Reddit again and again while talking about the paper, but looking at their design and Reddit’s front-page, the connection is obvious.

GroupLens
Reddit

Comparing them makes us realize where Reddit got their idea of User Interface.

To be honest, even though I was focusing mostly on the similarity of GroupLens and Reddit (as Reddit came to mind throughout the time i was reading the paper), we see the influence GroupLens has on overall modern Internet. The recommendation system, maybe not as simplified as before, is used by so many websites such as Youtube, Spotify, Amazon etc. They are used for advertisement, better recommendation of video, music or products. We can definitely see how each of features that GroupLens introduced is used these days throughout the internet and not just the netnews websites.